diff --git a/.config/nextest.toml b/.config/nextest.toml
index 8cba98c..693ab4a 100644
--- a/.config/nextest.toml
+++ b/.config/nextest.toml
@@ -1,6 +1,6 @@
# Nextest configuration — https://nexte.st/docs/configuration/
#
-# The point of this file is a per-test timeout safety net. Luxe's test *execution* is fast
+# The point of this file is a per-test timeout safety net. luxe's test *execution* is fast
# (~30s for the whole workspace once compiled), but a genuinely hung or runaway test — e.g. a
# process-spawning integration test that never reaps its child, or the historical review-gate
# fork bomb — could otherwise wedge the machine for many minutes. Nextest kills any test that
diff --git a/.gitattributes b/.gitattributes
index 6432485..41fc5ef 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -14,7 +14,7 @@
# Development-process artifacts — excluded from published release archives.
#
-# These are Luxe's own planning spine, journal, and dogfooding memory. Luxe never
+# These are luxe's own planning spine, journal, and dogfooding memory. luxe never
# *requires* them at runtime (every reader treats them as read-if-present), so a fresh
# consumer repo starts clean. `git archive` (and scripts/publish-clean.sh) drop them via
# the `export-ignore` attribute below; the files stay in this development repo for history.
diff --git a/.gitignore b/.gitignore
index a3f78a3..ff74a9f 100644
--- a/.gitignore
+++ b/.gitignore
@@ -21,7 +21,7 @@ __pycache__/
.DS_Store
Thumbs.db
-# Luxe-specific
+# luxe-specific
.luxe/
# Verification captures — throwaway evidence for one session, megabytes of it.
.pi/agent-loop/screenshots/
diff --git a/AGENTS.md b/AGENTS.md
index 236e252..8f2efae 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -1,6 +1,6 @@
# AGENTS.md
-Rules for working on Luxe: only what you'd get wrong without being told. It's sent with every
+Rules for working on luxe: only what you'd get wrong without being told. It's sent with every
request, so keep it short; the long version is [`CONTRIBUTING.md`](CONTRIBUTING.md).
## Commands
@@ -52,7 +52,7 @@ cargo run -p luxe-cli -- --headless -p "..." # one prompt, no UI
- A UI talks to core over channels only: `Action` in, `Event` out. It never calls a provider or a
tool and never blocks the runtime.
- Behavior that cuts across the loop is a `Hook`, not a special case in the loop.
-- One owner per platform question: `user_dirs` (home and Luxe's folders), `program` (finding
+- One owner per platform question: `user_dirs` (home and luxe's folders), `program` (finding
programs and shells, detaching, killing a process tree), `paths` (path to string), `fs_atomic`
(safe file replace), `luxe-lsp::uri` (`file://` URLs).
@@ -108,7 +108,7 @@ Enforce these in code (the permission engine or a hook), never in prompt text.
`gate.commands`, MCP and language servers need `--trust-project` (or global
`lsp.project_servers`), so "always" approvals save globally. `project_tooling_allowed` is the one
switch for project binaries; reuse it. Language servers still run `build.rs` and proc macros, so
- never say Luxe runs no repo code.
+ never say luxe runs no repo code.
- Secrets: `0600` files, `$ENV` or `!command`, held in `secrecy::SecretString`, scrubbed from tool
results, logs and wire captures by `redact::process_secrets()`.
- No telemetry. Every outbound call is one the user asked for or can turn off, sends nothing about
@@ -159,7 +159,7 @@ Enforce these in code (the permission engine or a hook), never in prompt text.
- Glyphs decorate; they never carry meaning alone. Every icon sits beside its words. Prefer block
elements, which every terminal draws.
- `web/src/assets/generated/` comes from Rust and CI checks it; `captures/` are real sessions from
- `scripts/site-record.sh`. Change how Luxe looks and you owe a `site-assets.sh` run and a
+ `scripts/site-record.sh`. Change how luxe looks and you owe a `site-assets.sh` run and a
re-record.
- The wordmark constants in `luxe-tui/src/widgets/wordmark/art.rs` are the logo: one drawing, no
alternatives.
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 35aa74d..d304310 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,7 +1,7 @@
# Changelog
-Every release of Luxe, newest first. The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
-and versions follow [Semantic Versioning](https://semver.org/spec/v2.0.0.html). While Luxe is below
+Every release of luxe, newest first. The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
+and versions follow [Semantic Versioning](https://semver.org/spec/v2.0.0.html). While luxe is below
1.0, a minor bump (0.x.0) may break config, flags or the `--rpc` protocol, and a patch bump never
does.
@@ -9,6 +9,43 @@ The version on `main` is always the next release. The first change to what ships
`Cargo.toml`, `Cargo.lock`) after a release bumps it one step and opens its section here, and later
changes add to that section until it ships; `scripts/version-check.py` holds them to it.
+## [0.1.4]
+
+### Added
+
+- luxe remembers the model you last chose in each project, and starts there. A project you haven't
+ chosen in yet starts on the model you last chose anywhere, then on `default_model`.
+- The mouse wheel and the scrollbar work in every list in a modal: the model picker, the palette,
+ themes, sessions, providers, settings and the `/` and `@` flyouts. Both move the cursor, so Enter
+ never picks a row you can't see.
+- The dock's telemetry row starts with the provider, beside its own mark where the font has one.
+
+### Changed
+
+- The name is written lowercase everywhere, as **luxe**, even to start a sentence. The dot in
+ **luxe.** belongs to the logo.
+- `luxe --help` opens with "A coding agent for your terminal."
+- Signing in no longer picks a model for you. The model picker opens at the top of its list, and
+ nothing changes until you choose.
+- Loop output marks each phase with a colored rail and each iteration with a bracket inside it. In
+ a terminal a tool's line shows `○` while it runs, then a green `✓` or a red `✗`.
+- The phone app shows a turn's reasoning in full, as the interface does. It was cut to three lines
+ with nothing to say there was more.
+- A phone notification for a question reads "The agent has a question for you".
+
+### Fixed
+
+- On Windows, `luxe update` replaces the running luxe. Windows won't overwrite a program while
+ it runs, so the old copy is moved aside and deleted at the next launch.
+- A luxe that's already signed in no longer opens first-run setup, even if setup was never
+ finished. Quitting it partway left nothing saved, so it came back at the next launch.
+- Opening a large project in the file tree no longer shows "engine busy". It asked for every
+ changed folder at once and overflowed the queue; it now asks for a few at a time.
+- The phone app stays where you are while you read back through a long session. Each new item
+ dropped the oldest from the top of the page, which moved everything you were reading.
+- A question answered from the phone app shows its answer on its card, as the interface does. It
+ was hidden behind a tap, as raw text.
+
## [0.1.3]
### Added
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 3426989..166430c 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -1,6 +1,6 @@
-# Contributing to Luxe
+# Contributing to luxe
-Process and tooling for people working on Luxe itself. The rules an agent has to hold to are in
+Process and tooling for people working on luxe itself. The rules an agent has to hold to are in
[`AGENTS.md`](AGENTS.md), deliberately kept short because it rides in every request; this file is
the long version, for humans, and nothing here is loaded into a prompt.
@@ -11,18 +11,18 @@ editing a page, and a test fails until you do.
## Licensing your contribution
-Luxe is dual licensed under MIT (`LICENSE-MIT`) or Apache-2.0 (`LICENSE-APACHE`), at the user's
+luxe is dual licensed under MIT (`LICENSE-MIT`) or Apache-2.0 (`LICENSE-APACHE`), at the user's
option. Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in
the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any
additional terms or conditions.
## AI assistance
-AI help is welcome. Luxe is built with plenty of it, much of it Luxe working on itself. Three things
+AI help is welcome. luxe is built with plenty of it, much of it luxe working on itself. Three things
to hold to:
- **Say so.** Name the tool and how much of the change it wrote in the pull request: "written with
- Luxe on Claude Sonnet 5.5, then reviewed and edited by me", or "asked ChatGPT how the cache works,
+ luxe on Claude Sonnet 5.5, then reviewed and edited by me", or "asked ChatGPT how the cache works,
wrote the fix myself". Tab completion doesn't count. A commit can carry the Linux kernel's
`Assisted-by: AGENT:MODEL` trailer.
- **Understand it.** You should be able to explain what your change does and how it fits without
@@ -280,7 +280,7 @@ python3 scripts/voice-check.py --review
The Tools page's bars are recorded, not generated: `scripts/tool-bench.sh NAME=REPO:CASES ...` runs
each tool and the shell command it replaces in each checkout you name (the cases are in
-`scripts/tool-bench/`), then in Luxe itself, and writes `web/src/data/tool-bench.json` naming every
+`scripts/tool-bench/`), then in luxe itself, and writes `web/src/data/tool-bench.json` naming every
repository's commit. Each shell row is also put through the other agents' own output limits, since
that, not the raw output, is what their models are sent. Nothing checks the file; run it on
committed trees when a tool's output changes. The page draws from the first repository.
@@ -327,6 +327,7 @@ npm ci # TypeScript, Playwright, and the vendo
npm run check # tsc
npm test # unit tests
cargo build -p luxe-cli && npm run e2e # end to end; LUXE_BIN picks another luxe
+node e2e/screenshots.mjs # the guide's phone shots, into docs/guide/phone/
```
Every iOS browser is WebKit, so check the app there. Playwright's WebKit build wants Ubuntu's ICU,
@@ -386,7 +387,7 @@ packages and builds every crate the way the registry will, dependencies first.
## Development-only files
-A development checkout carries planning files, design notes and evaluation rigs that Luxe never
+A development checkout carries planning files, design notes and evaluation rigs that luxe never
needs at runtime. They are marked `export-ignore` in `.gitattributes`, and
`scripts/publish-clean.sh
` writes a release tree without them, then scans what it wrote for
anything that must not ship.
diff --git a/Cargo.lock b/Cargo.lock
index 86e7c88..7568502 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1690,7 +1690,7 @@ checksum = "112b39cec0b298b6c1999fee3e31427f74f676e4cb9879ed1a121b43661a4154"
[[package]]
name = "luxe-cli"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"anyhow",
"async-trait",
@@ -1717,12 +1717,13 @@ dependencies = [
"tokio",
"tracing",
"tracing-subscriber",
+ "unicode-width",
"uuid",
]
[[package]]
name = "luxe-core"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"async-trait",
"base16ct",
@@ -1753,7 +1754,7 @@ dependencies = [
[[package]]
name = "luxe-dap"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"async-trait",
"luxe-core",
@@ -1767,7 +1768,7 @@ dependencies = [
[[package]]
name = "luxe-lsp"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"async-trait",
"base16ct",
@@ -1791,7 +1792,7 @@ dependencies = [
[[package]]
name = "luxe-mcp"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"async-trait",
"futures-util",
@@ -1811,7 +1812,7 @@ dependencies = [
[[package]]
name = "luxe-providers"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"async-trait",
"base64",
@@ -1830,7 +1831,7 @@ dependencies = [
[[package]]
name = "luxe-remote"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"axum",
"base16ct",
@@ -1853,7 +1854,7 @@ dependencies = [
[[package]]
name = "luxe-tools"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"anyhow",
"async-trait",
@@ -1879,7 +1880,7 @@ dependencies = [
[[package]]
name = "luxe-tui"
-version = "0.1.3"
+version = "0.1.4"
dependencies = [
"anyhow",
"arboard",
diff --git a/Cargo.toml b/Cargo.toml
index e7b588d..87ddd92 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -6,12 +6,12 @@ members = ["crates/*"]
# Every crate here releases together from one workspace, so the versions move in lockstep and each
# crate inherits this one. The internal `path` dependencies carry the same number explicitly — cargo
# uses the path locally and the version on crates.io, and refuses to publish a path dep without one.
-version = "0.1.3"
+version = "0.1.4"
edition = "2021"
license = "MIT OR Apache-2.0"
repository = "https://github.com/evanwhittaker/luxe"
homepage = "https://luxe.dev"
-# Bound by the dependencies: uuid needs 1.89, and ratatui, image and darling 1.88. Luxe's own floor
+# Bound by the dependencies: uuid needs 1.89, and ratatui, image and darling 1.88. luxe's own floor
# is lower (`user_dirs` relies on `std::env::home_dir`, which returned `$HOME` on Windows before
# 1.85). The `msrv` CI job builds the workspace on exactly this version, so the number cannot drift.
rust-version = "1.89"
@@ -106,7 +106,7 @@ image = { version = "0.25", default-features = false, features = [
[workspace.lints.rust]
# `deny`, not `forbid`, so an audited exception can carry its own `#[allow]`, which is the line a
# reviewer has to look at. The shipped program has one: `program::detach_from_terminal`, whose
-# `pre_exec(setsid)` is how a spawned command leaves Luxe's terminal (std's `Command::setsid` is not
+# `pre_exec(setsid)` is how a spawned command leaves luxe's terminal (std's `Command::setsid` is not
# stable). The others are the counting allocators in luxe-tui's frame-budget test and render-perf
# example, which never ship.
unsafe_code = "deny"
diff --git a/README.md b/README.md
index a5941d5..154a6db 100644
--- a/README.md
+++ b/README.md
@@ -1,15 +1,15 @@
-
+
-Luxe is a coding agent for your terminal, written in Rust. It works in your project with the model
+luxe is a coding agent for your terminal, written in Rust. It works in your project with the model
you choose, then reads back what it changed as a Guided Review, so you can comment, ask for fixes,
and open the pull request without leaving the terminal.
-
+
*One take on Claude Sonnet 5.5, with only the waits cut: it writes `todo overdue` with a test, the
change is read back in three chapters, and the pull request is drafted from the review.*
@@ -40,13 +40,13 @@ change is read back in three chapters, and the pull request is drafted from the
the turn fails, and `--format json` streams its messages, tool calls and each call's token usage.
With nobody to answer, a call that would ask is allowed; `--on-ask deny` refuses it instead.
`luxe --rpc` speaks the whole protocol.
-- **Extended with files and programs**, never code loaded into Luxe: skills, hooks, MCP servers,
- your own `@` sources, and an [RPC protocol](https://luxe.dev/docs/rpc/) for driving Luxe from
+- **Extended with files and programs**, never code loaded into luxe: skills, hooks, MCP servers,
+ your own `@` sources, and an [RPC protocol](https://luxe.dev/docs/rpc/) for driving luxe from
your own program.
-Luxe connects to 24 providers and any OpenAI-compatible endpoint. Sign in with a ChatGPT or GitHub
+luxe connects to 24 providers and any OpenAI-compatible endpoint. Sign in with a ChatGPT or GitHub
Copilot subscription or an OpenRouter account, or bring an API key; Anthropic and Google take API
-keys only. There is no telemetry: every connection Luxe makes is one you asked for or can turn off,
+keys only. There is no telemetry: every connection luxe makes is one you asked for or can turn off,
and [the guide lists them](https://luxe.dev/docs/security/#what-luxe-sends-over-the-network).

@@ -76,16 +76,16 @@ cargo build --release
./target/release/luxe --version
```
-Luxe runs on Linux, macOS and Windows, in any UTF-8 terminal; on Windows, use Windows Terminal and
+luxe runs on Linux, macOS and Windows, in any UTF-8 terminal; on Windows, use Windows Terminal and
install [Git for Windows](https://gitforwindows.org/) so the agent has a POSIX shell. You also need
a model to talk to: an API key or a subscription. Your settings and credentials live in
`~/.config/luxe/` (`%APPDATA%\luxe` on Windows).
Icons come from a [Nerd Font](https://www.nerdfonts.com/) when you have one, in its **Mono** build;
-without one, Luxe draws plain characters, and every icon has a word beside it. See
+without one, luxe draws plain characters, and every icon has a word beside it. See
[Fonts](https://luxe.dev/docs/install/#fonts).
-Luxe uses these when they are installed, and works without them:
+luxe uses these when they are installed, and works without them:
| | For |
| --- | --- |
@@ -95,7 +95,7 @@ Luxe uses these when they are installed, and works without them:
| A debug adapter per language | The `debug` tool |
| [ast-grep](https://ast-grep.github.io) | Search and replace by code structure |
-`luxe doctor` checks which of them Luxe can use on your machine, and how to install the rest.
+`luxe doctor` checks which of them luxe can use on your machine, and how to install the rest.
## Quickstart
@@ -110,7 +110,7 @@ Luxe uses these when they are installed, and works without them:
`luxe --worktree` does the same in a worktree, so your checkout is never touched.
2. **Answer setup.** The first launch asks for a theme, a provider, which tools the agent can use,
- and what Luxe opens on. Sign in with a subscription or paste an API key; a key already in your
+ and what luxe opens on. Sign in with a subscription or paste an API key; a key already in your
environment, such as `ANTHROPIC_API_KEY`, is found on its own. Choose a model when the picker
opens.
@@ -132,9 +132,9 @@ on the screen you are on, and `ctrl+c` quits; the session is already saved. The
## Documentation
-The guide is at [luxe.dev/docs](https://luxe.dev/docs/), and built into Luxe for the version you
+The guide is at [luxe.dev/docs](https://luxe.dev/docs/), and built into luxe for the version you
have: `luxe docs`, `luxe docs hooks`, `luxe docs --search "on_ask"`. Ask the agent how to do
-something in Luxe and it reads the answer from the same pages.
+something in luxe and it reads the answer from the same pages.
| To | Read |
| --- | --- |
@@ -148,7 +148,7 @@ something in Luxe and it reads the answer from the same pages.
| Give the agent your project's conventions | [Project context](https://luxe.dev/docs/project-context/) |
| Change the theme, keys or statusline | [Configuration](https://luxe.dev/docs/configuration/) |
| Control what the agent may do | [Security and permissions](https://luxe.dev/docs/security/) |
-| Add tools, skills, hooks or your own front end | [Extending Luxe](https://luxe.dev/docs/extending/) |
+| Add tools, skills, hooks or your own front end | [Extending luxe](https://luxe.dev/docs/extending/) |
| Look up a flag or an environment variable | [Command line](https://luxe.dev/docs/cli/), [Environment variables](https://luxe.dev/docs/environment/) |
| Fix something that is not working | [Troubleshooting](https://luxe.dev/docs/troubleshooting/) |
@@ -161,11 +161,11 @@ Every page is also plain markdown in [`docs/guide/`](docs/guide/), and
the site and the demos are made; [`AGENTS.md`](AGENTS.md) has the rules an agent working here holds
to. Problems and questions go to [GitHub issues](https://github.com/evanwhittaker/luxe/issues).
-Luxe is built with AI assistance, much of it Luxe working on itself. Every change still passes the
+luxe is built with AI assistance, much of it luxe working on itself. Every change still passes the
full test suite and CI before it merges, and its maintainer decides what does. If AI helped with
yours, [say so](CONTRIBUTING.md#ai-assistance).
-Luxe draws on Pi, by earendil-works, and on OpenCode, and on the unidirectional data flow of Elm
+luxe draws on Pi, by earendil-works, and on OpenCode, and on the unidirectional data flow of Elm
and Ratatui.
## License
diff --git a/SECURITY.md b/SECURITY.md
index 82f4fcf..1201161 100644
--- a/SECURITY.md
+++ b/SECURITY.md
@@ -1,13 +1,13 @@
# Security
-Luxe runs a model's commands on your machine, so a hole in it matters. If you find one, please tell
+luxe runs a model's commands on your machine, so a hole in it matters. If you find one, please tell
me privately before anyone else.
## Reporting
[Open a private advisory](https://github.com/evanwhittaker/luxe/security/advisories/new) on GitHub,
or email [contact@luxe.dev](mailto:contact@luxe.dev). Say what you did, what happened, and the
-Luxe version (`luxe --version`). A transcript or a minimal repository that shows it helps most.
+luxe version (`luxe --version`). A transcript or a minimal repository that shows it helps most.
I'll answer within a few days, keep you posted while it's fixed, and credit you in the advisory and
the release notes unless you'd rather not be named. Please give the fix time to ship before you
diff --git a/crates/luxe-cli/Cargo.toml b/crates/luxe-cli/Cargo.toml
index c30b8a3..173738a 100644
--- a/crates/luxe-cli/Cargo.toml
+++ b/crates/luxe-cli/Cargo.toml
@@ -3,7 +3,7 @@ name = "luxe-cli"
version.workspace = true
edition.workspace = true
license.workspace = true
-description = "A beautiful coding agent for your terminal."
+description = "A coding agent for your terminal."
repository.workspace = true
homepage.workspace = true
readme = "../../README.md"
@@ -18,14 +18,14 @@ name = "luxe"
path = "src/main.rs"
[dependencies]
-luxe-core = { path = "../luxe-core", version = "0.1.3" }
-luxe-providers = { path = "../luxe-providers", version = "0.1.3" }
-luxe-tools = { path = "../luxe-tools", version = "0.1.3" }
-luxe-lsp = { path = "../luxe-lsp", version = "0.1.3" }
-luxe-dap = { path = "../luxe-dap", version = "0.1.3" }
-luxe-mcp = { path = "../luxe-mcp", version = "0.1.3" }
-luxe-tui = { path = "../luxe-tui", version = "0.1.3" }
-luxe-remote = { path = "../luxe-remote", version = "0.1.3", optional = true }
+luxe-core = { path = "../luxe-core", version = "0.1.4" }
+luxe-providers = { path = "../luxe-providers", version = "0.1.4" }
+luxe-tools = { path = "../luxe-tools", version = "0.1.4" }
+luxe-lsp = { path = "../luxe-lsp", version = "0.1.4" }
+luxe-dap = { path = "../luxe-dap", version = "0.1.4" }
+luxe-mcp = { path = "../luxe-mcp", version = "0.1.4" }
+luxe-tui = { path = "../luxe-tui", version = "0.1.4" }
+luxe-remote = { path = "../luxe-remote", version = "0.1.4", optional = true }
# The pairing QR in the terminal. Without its defaults it is the encoder and text renderer alone,
# with no dependencies.
qrcode = { version = "0.14", default-features = false, optional = true }
@@ -48,18 +48,20 @@ reqwest = { workspace = true }
# `luxe auth login` reads an API key without echoing it, which needs raw mode on the real terminal.
# Pinned to luxe-tui's version so there is exactly one crossterm in the tree.
crossterm = "0.29"
+# The loop view cuts each line to the terminal's width, so it can redraw a line in place
+unicode-width = { workspace = true }
# Streaming the update manifest with a size cap, and comparing the version it reports.
futures = { workspace = true }
semver = { workspace = true }
[dev-dependencies]
tempfile = "3"
-# The guide check: what Luxe would silently ignore in a config example
+# The guide check: what luxe would silently ignore in a config example
serde_ignored = "0.1"
[features]
default = ["clipboard-images", "remote"]
-# `luxe remote`, the only server in Luxe. A packager can leave it out to ship a Luxe with no
+# `luxe remote`, the only server in luxe. A packager can leave it out to ship a luxe with no
# server code at all.
remote = ["dep:luxe-remote", "dep:qrcode"]
clipboard = ["luxe-tui/clipboard"]
diff --git a/crates/luxe-cli/examples/tool_bench.rs b/crates/luxe-cli/examples/tool_bench.rs
index 9a978a7..192f2a6 100644
--- a/crates/luxe-cli/examples/tool_bench.rs
+++ b/crates/luxe-cli/examples/tool_bench.rs
@@ -11,7 +11,7 @@
//! model is sent a raw 900 KB `git log`: the honest comparison is with what a harness forwards.
//!
//! `TOOL_BENCH_CASES` names a JSON file of cases (`id`, `shell`, `tool`, `args`) for measuring in a
-//! repository other than this one; without it, the cases below measure Luxe itself.
+//! repository other than this one; without it, the cases below measure luxe itself.
use std::process::Command;
use std::sync::Arc;
diff --git a/crates/luxe-cli/src/activity_projector/tests.rs b/crates/luxe-cli/src/activity_projector/tests.rs
index 7f9cab3..ab2065d 100644
--- a/crates/luxe-cli/src/activity_projector/tests.rs
+++ b/crates/luxe-cli/src/activity_projector/tests.rs
@@ -62,7 +62,7 @@ fn a_failed_call_summarizes_its_own_error() {
"edited before reading the file",
),
(
- "Error: 'src/lib.rs' exists and has no read on record for this Luxe run \
+ "Error: 'src/lib.rs' exists and has no read on record for this luxe run \
(read-before-write guard), so overwriting it would be blind.",
"overwrote a file it hadn't read",
),
diff --git a/crates/luxe-cli/src/cli.rs b/crates/luxe-cli/src/cli.rs
index c4fc572..bab837d 100644
--- a/crates/luxe-cli/src/cli.rs
+++ b/crates/luxe-cli/src/cli.rs
@@ -40,6 +40,44 @@ pub(crate) fn auth_service() -> luxe_tui::auth_wizard::AuthServiceHandle {
std::sync::Arc::new(luxe_providers::auth::ProviderAuth::new())
}
+/// Carries the flags that shape this run, rather than where it starts, onto `settings`.
+fn apply_run_flags(args: &Args, settings: &mut crate::engine_setup::AgentSettings) {
+ settings.resume = resume_from_args(args);
+ settings.permission_mode = permission_mode_from_args(args);
+ settings.trust_project = args.trust_project;
+ settings.disable_tools = args.no_tool.clone();
+ settings.enable_tools = args.with_tool.clone();
+ apply_thinking_flag(args.thinking.as_deref(), settings);
+ settings.auto_approve = args.auto_approve;
+ if let Some(on_ask) = args.on_ask {
+ settings.on_ask = on_ask.into();
+ }
+ if args.auto_approve && !args.prompts_a_human() && args.starts_an_engine() {
+ settings.startup_notices.push(
+ "--auto-approve: an unattended run already allows an `ask`. `--on-ask deny` is how to \
+ make one refuse instead."
+ .to_string(),
+ );
+ }
+ if args.plan {
+ settings.initial_mode = Some(luxe_core::mode::PLAN.to_string());
+ }
+}
+
+/// Pins the thinking level `--thinking` names; a level luxe doesn't know is reported and ignored.
+fn apply_thinking_flag(raw: Option<&str>, settings: &mut crate::engine_setup::AgentSettings) {
+ let Some(raw) = raw else {
+ return;
+ };
+ match args::parse_thinking_level(raw) {
+ Some(level) => {
+ settings.initial_thinking = Some(level);
+ settings.thinking_forced = true;
+ }
+ None => eprintln!("luxe: ignoring unknown --thinking level '{raw}'"),
+ }
+}
+
/// Top-level dispatch: parse args and route to the requested mode.
pub(crate) async fn run(mut args: Args) -> anyhow::Result<()> {
// The guide is in the binary and needs nothing else: no project, no settings, no network
@@ -58,34 +96,7 @@ pub(crate) async fn run(mut args: Args) -> anyhow::Result<()> {
let root = enter_project(&mut args, &invoked_from)?;
// Resolve provider/model/auth once (CLI flags > env > config) and share it downstream.
let mut settings = crate::engine_setup::AgentSettings::resolve(root, args.overrides());
- settings.resume = resume_from_args(&args);
- settings.permission_mode = permission_mode_from_args(&args);
- settings.trust_project = args.trust_project;
- settings.disable_tools = args.no_tool.clone();
- settings.enable_tools = args.with_tool.clone();
- if let Some(raw) = args.thinking.as_deref() {
- match args::parse_thinking_level(raw) {
- Some(level) => {
- settings.initial_thinking = Some(level);
- settings.thinking_forced = true;
- }
- None => eprintln!("luxe: ignoring unknown --thinking level '{raw}'"),
- }
- }
- settings.auto_approve = args.auto_approve;
- if let Some(on_ask) = args.on_ask {
- settings.on_ask = on_ask.into();
- }
- if args.auto_approve && !args.prompts_a_human() && args.starts_an_engine() {
- settings.startup_notices.push(
- "--auto-approve: an unattended run already allows an `ask`. `--on-ask deny` is how to \
- make one refuse instead."
- .to_string(),
- );
- }
- if args.plan {
- settings.initial_mode = Some(luxe_core::mode::PLAN.to_string());
- }
+ apply_run_flags(&args, &mut settings);
if args.starts_an_engine() {
enter_worktree(&args, &mut settings).await?;
@@ -104,9 +115,13 @@ pub(crate) async fn run(mut args: Args) -> anyhow::Result<()> {
// loop run needs the right context window just as much. Best-effort and non-blocking.
crate::model_catalog::spawn_refresh();
- // Ask, at most daily, whether a newer Luxe exists. The answer lands in a cache and is read by
+ // Ask, at most daily, whether a newer luxe exists. The answer lands in a cache and is read by
// the *next* launch's dashboard, so nothing here is on the path to a first token.
crate::update_check::spawn_check(&settings.state_root);
+ // The copy a Windows update moved aside; nothing to find anywhere else
+ if cfg!(windows) {
+ update_command::remove_leftover();
+ }
if let Some(cmd) = args.command {
match cmd {
diff --git a/crates/luxe-cli/src/cli/args.rs b/crates/luxe-cli/src/cli/args.rs
index 67e8013..fd1dada 100644
--- a/crates/luxe-cli/src/cli/args.rs
+++ b/crates/luxe-cli/src/cli/args.rs
@@ -8,7 +8,7 @@ mod permission_mode;
pub(crate) use commands::*;
pub(crate) use permission_mode::PermissionModeArg;
-/// Luxe's command line. `--help` shows the crate description from `Cargo.toml`, not this doc.
+/// luxe's command line. `--help` shows the crate description from `Cargo.toml`, not this doc.
#[derive(Parser, Debug)]
#[command(name = "luxe", version, about)]
pub(crate) struct Args {
@@ -17,7 +17,7 @@ pub(crate) struct Args {
pub(crate) headless: bool,
/// Speak the JSONL front-end protocol on stdio: wire actions in, events out. For editor
- /// plugins, alternative UIs, and anything driving Luxe as a session rather than a one-shot.
+ /// plugins, alternative UIs, and anything driving luxe as a session rather than a one-shot.
/// See `docs/guide/rpc.md`.
#[arg(long, conflicts_with = "headless")]
pub(crate) rpc: bool,
@@ -279,7 +279,7 @@ impl Args {
}
}
-/// Parse a `--thinking` level name into a `ThinkingLevel`, accepting the same spellings Luxe
+/// Parse a `--thinking` level name into a `ThinkingLevel`, accepting the same spellings luxe
/// displays (`off`..`ultra`, plus `max` as an alias for `ultra`). Returns `None` for unknown input.
pub(crate) fn parse_thinking_level(raw: &str) -> Option {
let name = raw.trim().to_ascii_lowercase();
diff --git a/crates/luxe-cli/src/cli/args/commands.rs b/crates/luxe-cli/src/cli/args/commands.rs
index 7cf3c19..68be462 100644
--- a/crates/luxe-cli/src/cli/args/commands.rs
+++ b/crates/luxe-cli/src/cli/args/commands.rs
@@ -7,13 +7,13 @@ use clap::{Parser, Subcommand};
pub(crate) enum SubCommand {
/// Sign in to a provider, check credentials, or sign out.
Auth(AuthArgs),
- /// Read Luxe's user guide, for this version.
+ /// Read luxe's user guide, for this version.
Docs(DocsArgs),
/// Check which language servers, debug adapters and tools work on this machine.
Doctor,
- /// Set Luxe up: a theme, a provider, the tools and the start screen.
+ /// Set luxe up: a theme, a provider, the tools and the start screen.
Init(InitArgs),
- /// Join a session another Luxe on this machine is running, speaking the `--rpc` protocol on
+ /// Join a session another luxe on this machine is running, speaking the `--rpc` protocol on
/// stdin and stdout.
Join(JoinArgs),
/// Serve your sessions to your own phone or tablet, through Tailscale.
@@ -27,7 +27,7 @@ pub(crate) enum SubCommand {
Models(ModelsArgs),
/// List this project's sessions.
Sessions(SessionsArgs),
- /// Update Luxe with the command for however it was installed.
+ /// Update luxe with the command for however it was installed.
Update(UpdateArgs),
/// Summarize a `LUXE_WIRE_LOG` capture.
Wire(WireArgs),
@@ -109,10 +109,10 @@ pub(crate) struct ModelsArgs {
pub(crate) command: ModelsCommand,
}
-/// Models subcommand — what Luxe knows about a model, and where it learned it.
+/// Models subcommand — what luxe knows about a model, and where it learned it.
#[derive(Subcommand, Debug)]
pub(crate) enum ModelsCommand {
- /// List known models for the providers Luxe can route to.
+ /// List known models for the providers luxe can route to.
List {
/// Only this provider.
#[arg(long)]
@@ -148,7 +148,7 @@ pub(crate) struct InitArgs {
/// Join arguments.
#[derive(Parser, Debug)]
pub(crate) struct JoinArgs {
- /// The session: its id or the start of it, or the process id of the Luxe running it. Omit
+ /// The session: its id or the start of it, or the process id of the luxe running it. Omit
/// it to join the one running in this folder.
pub(crate) target: Option,
/// List the sessions running on this machine that can be joined, and exit.
diff --git a/crates/luxe-cli/src/cli/auth_command.rs b/crates/luxe-cli/src/cli/auth_command.rs
index 776ccf2..6992480 100644
--- a/crates/luxe-cli/src/cli/auth_command.rs
+++ b/crates/luxe-cli/src/cli/auth_command.rs
@@ -105,17 +105,11 @@ fn report_success(service: &ProviderAuth, outcome: &LoginOutcome) {
field("models", note);
}
field("saved to", &luxe_core::paths::home_relative(&service.store_path()));
- if let Some(model) = &outcome.suggested_model {
- field("model", &format!("{model} {}", style::dim("(now the default)")));
- }
// Advice sits apart from the summary: it's about what to do later, not what just happened.
for advice in &outcome.advice {
outln!("\n {}", style::dim(advice));
}
- match outcome.suggested_model {
- Some(_) => outln!("\nRun `luxe` to start, or `/model` to pick another."),
- None => outln!("\nRun `luxe` to start, then `/model` to pick a model."),
- }
+ outln!("\nRun `luxe` to start, then `/model` to pick a model.");
}
/// `luxe auth list` — every provider and how it's configured.
diff --git a/crates/luxe-cli/src/cli/commands.rs b/crates/luxe-cli/src/cli/commands.rs
index 8bbc947..d0bfac1 100644
--- a/crates/luxe-cli/src/cli/commands.rs
+++ b/crates/luxe-cli/src/cli/commands.rs
@@ -172,7 +172,7 @@ async fn start_loop(
event_rx,
running_rx,
kickoff,
- crate::loop_view::LoopView::new(quiet),
+ crate::loop_view::LoopView::new(quiet, phase),
)
.await?;
Ok(())
diff --git a/crates/luxe-cli/src/cli/docs_command.rs b/crates/luxe-cli/src/cli/docs_command.rs
index f086736..8c33c28 100644
--- a/crates/luxe-cli/src/cli/docs_command.rs
+++ b/crates/luxe-cli/src/cli/docs_command.rs
@@ -2,7 +2,7 @@
//!
//! The guide ships in the binary, so it works offline and always describes the version installed:
//! `scripts/guide-bundle.sh` writes every page of `docs/guide/` into `src/guide.md`, and this reads
-//! that bundle. The agent is told the command exists, so asking Luxe how to configure Luxe gets an
+//! that bundle. The agent is told the command exists, so asking luxe how to configure luxe gets an
//! answer from the manual rather than a guess.
use crate::cli::args::DocsArgs;
@@ -11,7 +11,7 @@ use crate::cli::args::DocsArgs;
const BUNDLE: &str = include_str!("../guide.md");
/// The sidebar's groups, in reading order. `web/src/content.config.ts` holds the same list.
-const GROUPS: [&str; 5] = ["Get started", "Using Luxe", "Customize", "Extend", "Reference"];
+const GROUPS: [&str; 5] = ["Get started", "Using luxe", "Customize", "Extend", "Reference"];
/// The most lines `--search` prints, so a common word cannot flood a tool result.
const SEARCH_LIMIT: usize = 40;
@@ -74,7 +74,7 @@ pub(crate) fn run(args: &DocsArgs) -> anyhow::Result<()> {
/// Every page, under its group, with its summary and how to read it.
fn index(pages: &[Page]) -> String {
let mut out = String::from(
- "Luxe's user guide, for this version. `luxe docs ` prints a page; \
+ "luxe's user guide, for this version. `luxe docs ` prints a page; \
`luxe docs --search ` finds lines across all of them.\n",
);
for group in GROUPS {
diff --git a/crates/luxe-cli/src/cli/doctor_command.rs b/crates/luxe-cli/src/cli/doctor_command.rs
index e0c5dcf..b7f5223 100644
--- a/crates/luxe-cli/src/cli/doctor_command.rs
+++ b/crates/luxe-cli/src/cli/doctor_command.rs
@@ -1,6 +1,6 @@
//! `luxe doctor` — is the tooling on this machine actually going to work?
//!
-//! Luxe launches language servers and debug adapters; it does not install them, and it does not
+//! luxe launches language servers and debug adapters; it does not install them, and it does not
//! bundle them: installation would mean owning each language's package manager and update policy.
//! A broken or absent one can otherwise stay invisible until a tool call comes back empty.
//!
@@ -63,7 +63,7 @@ async fn language_servers() {
let configured: std::collections::HashSet = user.keys().cloned().collect();
for language in registry.merge_patches(user) {
say(&format!(
- " FAIL {language:<12} lsp-servers.json gives no `command`, and Luxe ships none"
+ " FAIL {language:<12} lsp-servers.json gives no `command`, and luxe ships none"
));
}
let mut languages: Vec<(&String, &ServerEntry)> = registry.iter().collect();
diff --git a/crates/luxe-cli/src/cli/join_command.rs b/crates/luxe-cli/src/cli/join_command.rs
index a7070e2..97eb1f4 100644
--- a/crates/luxe-cli/src/cli/join_command.rs
+++ b/crates/luxe-cli/src/cli/join_command.rs
@@ -1,6 +1,6 @@
-//! `luxe join` — a session another Luxe is running, over stdin and stdout.
+//! `luxe join` — a session another luxe is running, over stdin and stdout.
//!
-//! The running Luxe shares its session on a local endpoint (`cli::share`), and this command is the
+//! The running luxe shares its session on a local endpoint (`cli::share`), and this command is the
//! plain way in: it picks the session from the records in the runtime folder, connects, and copies
//! bytes both ways. What crosses is the `--rpc` protocol untouched, so a client written for
//! `luxe --rpc` drives a joined session by launching `luxe join` instead.
@@ -63,7 +63,7 @@ fn nothing_to_join(running: &[LiveSession], target: Option<&str>, cwd: &str) ->
};
if running.is_empty() {
return format!(
- "{wanted}, and none is running on this machine that can be joined. A Luxe shares \
+ "{wanted}, and none is running on this machine that can be joined. A luxe shares \
its session unless `joinable` is off or LUXE_JOINABLE=0."
);
}
diff --git a/crates/luxe-cli/src/cli/join_command/tests.rs b/crates/luxe-cli/src/cli/join_command/tests.rs
index cb889d6..0ba4797 100644
--- a/crates/luxe-cli/src/cli/join_command/tests.rs
+++ b/crates/luxe-cli/src/cli/join_command/tests.rs
@@ -21,7 +21,7 @@ fn sessions() -> Vec {
/// With no target, the session running in this folder is the one joined.
///
/// The break this catches: ignoring the folder, so `luxe join` in one project lands in another's
-/// session, or refuses because more than one Luxe is running somewhere.
+/// session, or refuses because more than one luxe is running somewhere.
#[test]
fn no_target_joins_the_session_running_here() {
let all = sessions();
@@ -39,7 +39,7 @@ fn two_sessions_in_the_folder_are_listed_not_guessed() {
assert!(error.contains("aaab-2222") && error.contains("cccc-3333"), "{error}");
}
-/// A target names a session by the start of its id or by its Luxe's process id.
+/// A target names a session by the start of its id or by its luxe's process id.
///
/// The break this catches: matching the id exactly, so the short form the listing invites fails,
/// or not accepting a pid, which is what `ps` and herdr show.
@@ -62,7 +62,7 @@ fn an_ambiguous_or_empty_target_is_refused() {
/// Nothing to join says why, and how sharing is turned back on when nothing is running.
///
-/// The break this catches: a bare "not found" when every Luxe has `joinable` off, which leaves the
+/// The break this catches: a bare "not found" when every luxe has `joinable` off, which leaves the
/// reader with nothing to try.
#[test]
fn nothing_to_join_says_what_is_running() {
diff --git a/crates/luxe-cli/src/cli/models_command.rs b/crates/luxe-cli/src/cli/models_command.rs
index 88e59ef..8a2c3a8 100644
--- a/crates/luxe-cli/src/cli/models_command.rs
+++ b/crates/luxe-cli/src/cli/models_command.rs
@@ -1,6 +1,6 @@
//! `luxe models` — inspect and refresh the model catalog.
//!
-//! Exists mostly for one question: *why does Luxe think my model has a 128k window?* `show`
+//! Exists mostly for one question: *why does luxe think my model has a 128k window?* `show`
//! answers it directly by printing which layer won each field, so a stale cache, a missing
//! provider mapping, and a forgotten config override are all distinguishable at a glance.
diff --git a/crates/luxe-cli/src/cli/remote_command.rs b/crates/luxe-cli/src/cli/remote_command.rs
index 18a844b..d08d445 100644
--- a/crates/luxe-cli/src/cli/remote_command.rs
+++ b/crates/luxe-cli/src/cli/remote_command.rs
@@ -46,7 +46,7 @@ async fn serve(port: u16, allow: Vec, devices: DeviceBook) -> anyhow::Re
let logins = allowed_logins(tailnet, allow)?;
let access = Access { logins: logins.clone(), devices };
let bridge = bridge(port, access, machine).await?;
- outln!("Luxe remote is listening on http://{}. Ctrl-C stops it.", bridge.address());
+ outln!("luxe remote is listening on http://{}. Ctrl-C stops it.", bridge.address());
outln!("It lets in {}, from a paired device.", logins.join(", "));
bridge
.serve(async {
diff --git a/crates/luxe-cli/src/cli/remote_command/setup.rs b/crates/luxe-cli/src/cli/remote_command/setup.rs
index 0401641..911d87f 100644
--- a/crates/luxe-cli/src/cli/remote_command/setup.rs
+++ b/crates/luxe-cli/src/cli/remote_command/setup.rs
@@ -1,5 +1,5 @@
//! `luxe remote setup` — from nothing to a paired phone, asking before each step that changes
-//! something outside Luxe.
+//! something outside luxe.
//!
//! Four steps: Tailscale is there and signed in; Serve carries `https:///` to the
//! bridge; a user service runs the bridge at login; and an invitation to scan. Each step that is
@@ -108,7 +108,7 @@ fn service_step(port: u16, tailnet: &Tailnet, allow: Vec) {
return;
}
if !confirm("Run the bridge whenever you're signed in, as a user service?") {
- outln!("Skipped. Run `luxe remote` yourself whenever your devices should reach Luxe.");
+ outln!("Skipped. Run `luxe remote` yourself whenever your devices should reach luxe.");
return;
}
let launch =
diff --git a/crates/luxe-cli/src/cli/rpc/translate.rs b/crates/luxe-cli/src/cli/rpc/translate.rs
index c2562a3..7d7b8ae 100644
--- a/crates/luxe-cli/src/cli/rpc/translate.rs
+++ b/crates/luxe-cli/src/cli/rpc/translate.rs
@@ -75,7 +75,7 @@ impl GithubWrites {
match self {
Self::Allowed => None,
Self::NotEnabled => Some(format!(
- "{kind} publishes to GitHub, which this session does not allow: start Luxe with \
+ "{kind} publishes to GitHub, which this session does not allow: start luxe with \
--github-writes to let a client do that"
)),
Self::ReadOnly => Some(format!(
@@ -151,7 +151,7 @@ fn session_refusal(wire: &WireAction, policy: &Policy) -> Option {
let switches = matches!(wire, WireAction::NewSession | WireAction::LoadSession(_));
if policy.joined && switches {
let reason = "would switch the session the terminal is showing: switch it there, or start \
- another Luxe";
+ another luxe";
return Some(format!("{kind} {reason}"));
}
None
diff --git a/crates/luxe-cli/src/cli/rpc/translate/tests.rs b/crates/luxe-cli/src/cli/rpc/translate/tests.rs
index e6b320e..04771eb 100644
--- a/crates/luxe-cli/src/cli/rpc/translate/tests.rs
+++ b/crates/luxe-cli/src/cli/rpc/translate/tests.rs
@@ -447,7 +447,7 @@ fn github_writes_need_the_flag_and_read_only_refuses_them_anyway() {
]
};
for (gate, needle) in [
- (GithubWrites::NotEnabled, "start Luxe with --github-writes"),
+ (GithubWrites::NotEnabled, "start luxe with --github-writes"),
(GithubWrites::ReadOnly, "read-only"),
] {
for wire in writes() {
diff --git a/crates/luxe-cli/src/cli/tui.rs b/crates/luxe-cli/src/cli/tui.rs
index 8b8b446..dc482fd 100644
--- a/crates/luxe-cli/src/cli/tui.rs
+++ b/crates/luxe-cli/src/cli/tui.rs
@@ -11,7 +11,7 @@ fn footer_note(repo_root: &std::path::Path) -> Option {
crate::update_check::cached_notice(repo_root).map(|_| UPDATE_HINT.to_string())
}
-/// What the dashboard says when a newer Luxe exists.
+/// What the dashboard says when a newer luxe exists.
const UPDATE_HINT: &str = "Update available · /update";
/// Whether the startup dashboard should be shown.
@@ -295,7 +295,7 @@ async fn starting_status(
/// mid-session shows up without relaunching. The probe is the same menu built after asking every
/// connected local server what it is serving — a list that changes the moment somebody loads a
/// model, and that only the endpoint knows. An Azure resource is the same question asked of a
-/// cloud: what is deployed there changes in a portal Luxe never sees. A server that is off leaves
+/// cloud: what is deployed there changes in a portal luxe never sees. A server that is off leaves
/// the record standing.
fn model_menu_sources(
settings: &crate::engine_setup::AgentSettings,
diff --git a/crates/luxe-cli/src/cli/update_command.rs b/crates/luxe-cli/src/cli/update_command.rs
index 53593c7..4121a53 100644
--- a/crates/luxe-cli/src/cli/update_command.rs
+++ b/crates/luxe-cli/src/cli/update_command.rs
@@ -1,8 +1,8 @@
//! `luxe update` — run the update this install's owner provides, then check that it took.
//!
//! The policy is [`luxe_core::update`]'s: which owner installed this copy, which commands update
-//! it, and a refusal where Luxe cannot know them. This runs those commands one program at a time
-//! with the terminal attached, so the reader sees the package manager's own output, and Luxe never
+//! it, and a refusal where luxe cannot know them. This runs those commands one program at a time
+//! with the terminal attached, so the reader sees the package manager's own output, and luxe never
//! writes a binary itself. `/update` on the dashboard comes here too, once the TUI has handed the
//! terminal back.
@@ -10,6 +10,8 @@ use anyhow::{anyhow, bail, Context};
use luxe_core::update::{self, InstallSource};
use std::path::{Path, PathBuf};
+pub(crate) use aside::remove_leftover;
+
/// Checks for a newer release now and, unless `check_only`, runs the update and checks the result.
pub(crate) async fn run(check_only: bool) -> anyhow::Result<()> {
let current = luxe_core::version::VERSION;
@@ -17,50 +19,62 @@ pub(crate) async fn run(check_only: bool) -> anyhow::Result<()> {
.await
.map_err(|error| anyhow!("could not check for a newer release: {error}"))?;
if !update::is_newer(&latest, current) {
- println!("Luxe {current} is the newest release.");
+ println!("luxe {current} is the newest release.");
return Ok(());
}
- println!("Luxe {latest} is available (you have {current}).");
+ println!("luxe {latest} is available (you have {current}).");
let source = InstallSource::detect();
match (source.update_steps(), check_only) {
- (Some(steps), false) if !cfg!(windows) => {
- run_steps(steps).await?;
- report(installed_now(&latest).await)
- }
+ (Some(steps), false) => update_with(steps, &latest).await,
_ => {
- println!("{}", instructions(source, check_only, cfg!(windows)));
+ println!("{}", instructions(source, check_only));
Ok(())
}
}
}
-/// What to tell a reader Luxe will not run an update for: they asked only to check, Windows cannot
-/// replace the running program, or no owner's command is known.
-fn instructions(source: InstallSource, check_only: bool, windows: bool) -> String {
+/// Runs the owner's steps over this copy, then checks the `luxe` a shell finds now.
+///
+/// Windows won't let a running program's file be overwritten, and `luxe update` is itself a
+/// running luxe. It does allow a rename, so the running copy steps aside first and the owner
+/// writes a new file where it was.
+async fn update_with(steps: &[&[&str]], latest: &str) -> anyhow::Result<()> {
+ let exe = std::env::current_exe().context("could not find the running luxe")?;
+ let aside = match cfg!(windows) {
+ true => Some(aside::MovedAside::move_running(&exe)?),
+ false => None,
+ };
+ let ran = run_steps(steps, exe.parent()).await;
+ if let Some(aside) = &aside {
+ aside.restore_if_missing()?;
+ }
+ ran?;
+ report(installed_now(latest, &exe).await)
+}
+
+/// What to tell a reader luxe will not run an update for: they asked only to check, or no owner's
+/// command is known.
+fn instructions(source: InstallSource, check_only: bool) -> String {
match source.update_command() {
- Some(command) if !check_only && windows => format!(
- "Windows can't replace Luxe while it runs, so run this once it has exited: {command}"
- ),
Some(_) if check_only => format!("Run `luxe update`, or {}.", source.advice()),
_ => format!("To update: {}.", source.advice()),
}
}
-/// Runs each step in turn, the terminal attached, stopping at the first that fails.
-async fn run_steps(steps: &[&[&str]]) -> anyhow::Result<()> {
+/// Runs each step in turn, the terminal attached, stopping at the first that fails. `beside` is
+/// the running binary's directory.
+async fn run_steps(steps: &[&[&str]], beside: Option<&Path>) -> anyhow::Result<()> {
for step in steps {
let Some((program, args)) = step.split_first() else { continue };
let line = step.join(" ");
println!("Running: {line}");
- let beside =
- std::env::current_exe().ok().and_then(|exe| exe.parent().map(Path::to_path_buf));
- let status = tokio::process::Command::new(program_path(program, beside.as_deref()))
+ let status = tokio::process::Command::new(program_path(program, beside))
.args(args)
.status()
.await
.with_context(|| format!("could not start `{program}`"))?;
if !status.success() {
- bail!("`{line}` failed ({status}), so Luxe is unchanged");
+ bail!("`{line}` failed ({status}), so luxe is unchanged");
}
}
Ok(())
@@ -108,15 +122,15 @@ fn outcome(found: Option<&Path>, updated: &Path, reported: Option<&str>, latest:
}
}
-/// The `luxe` a shell would run now, and the outcome it shows.
-async fn installed_now(latest: &str) -> Outcome {
+/// The `luxe` a shell would run now, and the outcome it shows. `updated` is where the running copy
+/// was, read before a Windows update moved it aside.
+async fn installed_now(latest: &str, updated: &Path) -> Outcome {
let found = luxe_core::program::resolve("luxe");
let reported = match &found {
Some(path) => version_of(path).await,
None => None,
};
- let updated = std::env::current_exe().unwrap_or_default();
- outcome(found.as_deref(), &updated, reported.as_deref(), latest)
+ outcome(found.as_deref(), updated, reported.as_deref(), latest)
}
/// The version `path --version` prints, as `luxe 0.2.0`.
@@ -150,5 +164,7 @@ fn report(outcome: Outcome) -> anyhow::Result<()> {
Ok(())
}
+mod aside;
+
#[cfg(test)]
mod tests;
diff --git a/crates/luxe-cli/src/cli/update_command/aside.rs b/crates/luxe-cli/src/cli/update_command/aside.rs
new file mode 100644
index 0000000..2697f43
--- /dev/null
+++ b/crates/luxe-cli/src/cli/update_command/aside.rs
@@ -0,0 +1,118 @@
+//! Moving the running copy aside — how `luxe update` gets out of its own way on Windows.
+//!
+//! Windows locks a running program's file against being overwritten or deleted, but not against
+//! being renamed. The running `luxe.exe` becomes `luxe.exe.old`, the owner's updater writes a new
+//! `luxe.exe`, and the next launch deletes the leftover once nothing is running from it. Nothing
+//! here writes a binary: it only renames the one that is already there.
+
+use anyhow::Context;
+use std::path::{Path, PathBuf};
+
+/// The running copy, renamed out of the way of the update.
+pub(super) struct MovedAside {
+ from: PathBuf,
+ to: PathBuf,
+}
+
+impl MovedAside {
+ /// Renames `exe` to its leftover name, clearing a leftover a previous update left behind.
+ pub(super) fn move_running(exe: &Path) -> anyhow::Result {
+ let to = leftover_path(exe);
+ remove_if_present(&to).with_context(|| {
+ format!("could not remove {}, left by an earlier update", to.display())
+ })?;
+ std::fs::rename(exe, &to).with_context(|| {
+ format!("could not move the running {} aside to update it", exe.display())
+ })?;
+ Ok(Self { from: exe.to_path_buf(), to })
+ }
+
+ /// Puts the running copy back when the update left nothing where it was, so a failed update
+ /// never leaves the reader without a `luxe`.
+ pub(super) fn restore_if_missing(&self) -> anyhow::Result<()> {
+ if self.from.exists() {
+ return Ok(());
+ }
+ std::fs::rename(&self.to, &self.from).with_context(|| {
+ format!(
+ "the update left no luxe behind and {} could not be moved back to {}",
+ self.to.display(),
+ self.from.display()
+ )
+ })
+ }
+}
+
+/// Deletes the copy an earlier update moved aside, if there is one. Called at launch, when the
+/// old copy has exited. Still running elsewhere is normal (another session), so a failure is
+/// only logged and the next launch tries again.
+pub(crate) fn remove_leftover() {
+ let Ok(exe) = std::env::current_exe() else { return };
+ let leftover = leftover_path(&exe);
+ if let Err(error) = remove_if_present(&leftover) {
+ tracing::debug!(%error, path = %leftover.display(), "leftover from an update is still in use");
+ }
+}
+
+/// `luxe.exe` beside itself as `luxe.exe.old`.
+fn leftover_path(exe: &Path) -> PathBuf {
+ let mut name = exe.file_name().unwrap_or_default().to_os_string();
+ name.push(".old");
+ exe.with_file_name(name)
+}
+
+/// Removes `path`; already gone is success.
+fn remove_if_present(path: &Path) -> std::io::Result<()> {
+ match std::fs::remove_file(path) {
+ Err(error) if error.kind() == std::io::ErrorKind::NotFound => Ok(()),
+ other => other,
+ }
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ fn installed() -> (tempfile::TempDir, PathBuf) {
+ let dir = tempfile::tempdir().expect("tempdir");
+ let exe = dir.path().join("luxe.exe");
+ std::fs::write(&exe, b"old").expect("a stand-in binary");
+ (dir, exe)
+ }
+
+ #[test]
+ fn the_leftover_sits_beside_the_binary() {
+ assert_eq!(leftover_path(Path::new("/bin/luxe.exe")), PathBuf::from("/bin/luxe.exe.old"));
+ }
+
+ /// The update writes a new file where the old one was, and the old one is left to delete.
+ #[test]
+ fn an_update_that_wrote_a_new_copy_keeps_it() {
+ let (_dir, exe) = installed();
+ let aside = MovedAside::move_running(&exe).expect("moved");
+ assert!(!exe.exists(), "the name is free for the updater");
+ std::fs::write(&exe, b"new").expect("the updater's copy");
+ aside.restore_if_missing().expect("nothing to restore");
+ assert_eq!(std::fs::read(&exe).expect("read"), b"new");
+ assert_eq!(std::fs::read(leftover_path(&exe)).expect("read"), b"old");
+ }
+
+ /// A failed update puts the running copy back rather than leaving no `luxe` at all.
+ #[test]
+ fn an_update_that_wrote_nothing_puts_the_old_copy_back() {
+ let (_dir, exe) = installed();
+ let aside = MovedAside::move_running(&exe).expect("moved");
+ aside.restore_if_missing().expect("restored");
+ assert_eq!(std::fs::read(&exe).expect("read"), b"old");
+ assert!(!leftover_path(&exe).exists());
+ }
+
+ /// A leftover from the last update is cleared, not a reason to refuse this one.
+ #[test]
+ fn a_leftover_from_last_time_is_replaced() {
+ let (_dir, exe) = installed();
+ std::fs::write(leftover_path(&exe), b"older").expect("a leftover");
+ MovedAside::move_running(&exe).expect("moved");
+ assert_eq!(std::fs::read(leftover_path(&exe)).expect("read"), b"old");
+ }
+}
diff --git a/crates/luxe-cli/src/cli/update_command/tests.rs b/crates/luxe-cli/src/cli/update_command/tests.rs
index f441320..fe3ac0c 100644
--- a/crates/luxe-cli/src/cli/update_command/tests.rs
+++ b/crates/luxe-cli/src/cli/update_command/tests.rs
@@ -69,13 +69,8 @@ fn no_luxe_on_path_or_no_version_cannot_be_judged() {
/// Only the reader's own command is ever named, and a check offers `luxe update` beside it.
#[test]
fn the_instructions_name_the_owners_command_or_none() {
- let check = instructions(InstallSource::Cargo, true, false);
+ let check = instructions(InstallSource::Cargo, true);
assert!(check.contains("luxe update") && check.contains("cargo install"), "{check}");
- let download = instructions(InstallSource::Archive, false, false);
+ let download = instructions(InstallSource::Archive, false);
assert!(download.contains("download it from"), "{download}");
- let windows = instructions(InstallSource::Cargo, false, true);
- assert!(
- windows.contains("once it has exited") && windows.contains("cargo install"),
- "{windows}"
- );
}
diff --git a/crates/luxe-cli/src/cli/wire_command/secrets.rs b/crates/luxe-cli/src/cli/wire_command/secrets.rs
index 8284548..5eca0d1 100644
--- a/crates/luxe-cli/src/cli/wire_command/secrets.rs
+++ b/crates/luxe-cli/src/cli/wire_command/secrets.rs
@@ -7,7 +7,7 @@ const SECRET_PREFIXES: &[&str] = &["sk-", "sk_live_", "ghp_", "gho_", "xoxb-", "
/// Minimum run of key-shaped characters after a prefix before it counts as a credential.
///
-/// Without this, reading Luxe's own `redact.rs` — whose source *contains* every prefix — reports a
+/// Without this, reading luxe's own `redact.rs` — whose source *contains* every prefix — reports a
/// leak on every audit. The prefix alone is prose; the prefix plus a key is a secret.
const SECRET_BODY: usize = 16;
diff --git a/crates/luxe-cli/src/cli/wire_command/tests.rs b/crates/luxe-cli/src/cli/wire_command/tests.rs
index cf2ce53..a66b28a 100644
--- a/crates/luxe-cli/src/cli/wire_command/tests.rs
+++ b/crates/luxe-cli/src/cli/wire_command/tests.rs
@@ -184,7 +184,7 @@ fn a_bold_label_outside_the_tools_section_is_not_a_duplicate() {
#[test]
fn a_credential_prefix_in_prose_is_not_a_leak_but_a_real_key_is() {
- // Reading Luxe's own redact.rs puts every prefix in a tool result; that is not a leak.
+ // Reading luxe's own redact.rs puts every prefix in a tool result; that is not a leak.
assert!(looks_like_credential("patterns like sk- and ghp_ and AKIA").is_none());
assert!(looks_like_credential("Authorization: Bearer short").is_none());
let leak = looks_like_credential("key=sk-abcdefghijklmnopqrstuvwxyz0123")
diff --git a/crates/luxe-cli/src/cli/worktree.rs b/crates/luxe-cli/src/cli/worktree.rs
index d8c847e..bcf7486 100644
--- a/crates/luxe-cli/src/cli/worktree.rs
+++ b/crates/luxe-cli/src/cli/worktree.rs
@@ -248,7 +248,7 @@ pub(super) fn worktree_notice(
)
}
-/// Leaves the session's worktree and removes it, saying how it went on the terminal Luxe has
+/// Leaves the session's worktree and removes it, saying how it went on the terminal luxe has
/// just handed back.
pub(super) async fn remove_worktree(project: &std::path::Path, path: &std::path::Path) {
if let Err(error) = std::env::set_current_dir(project) {
diff --git a/crates/luxe-cli/src/config_editor.rs b/crates/luxe-cli/src/config_editor.rs
index 9975929..611d089 100644
--- a/crates/luxe-cli/src/config_editor.rs
+++ b/crates/luxe-cli/src/config_editor.rs
@@ -1,4 +1,4 @@
-//! Config editor handoff — opens one of Luxe's files outside the alternate screen.
+//! Config editor handoff — opens one of luxe's files outside the alternate screen.
//!
//! The TUI returns control before this runs, so terminal editors inherit a normal terminal.
//! When the editor exits, the CLI rebuilds the TUI and reloads the updated configuration.
@@ -47,7 +47,7 @@ pub(crate) async fn open_config_file(path: PathBuf) -> anyhow::Result {
/// credential file created world-readable is a credential file leaked.
fn ensure_file(path: PathBuf) -> anyhow::Result {
if path == Config::global_path() {
- return Config::ensure_global_file().context("creating Luxe user config");
+ return Config::ensure_global_file().context("creating luxe user config");
}
if path.exists() {
return Ok(path);
diff --git a/crates/luxe-cli/src/engine_setup.rs b/crates/luxe-cli/src/engine_setup.rs
index 3fb6ac9..5909b7a 100644
--- a/crates/luxe-cli/src/engine_setup.rs
+++ b/crates/luxe-cli/src/engine_setup.rs
@@ -135,7 +135,7 @@ pub(crate) async fn build_engine(
let model = settings.launch_model();
- // Provider routing: the adapter the settings selected, plus a registry of every provider Luxe
+ // Provider routing: the adapter the settings selected, plus a registry of every provider luxe
// knows about, so an in-session model switch *or* a mid-session `/login` reaches the right one.
let provider = build_active_adapter(settings);
let providers = build_provider_registry(settings, &provider);
@@ -183,8 +183,9 @@ pub(crate) async fn build_engine(
external_work: Some(services::shared_external_work()),
review_context_files: config.review.context_files.clone(),
review_exclude: config.review.exclude.clone(),
- // A real session remembers what the user picked; embedders and tests leave no trace.
- remember_runtime_state: config.agent.remember_model(),
+ // A real session remembers what the user picked, here and anywhere; embedders and tests
+ // leave no trace.
+ remember_choices_in: config.agent.remember_model().then(|| settings.work_root.clone()),
cache_warming: config.agent.cache_warming(),
// Ground Guided Review in real structure (relevant files + test gaps) from the warm graph.
review_insight_provider: Some(Arc::new(crate::review_insight::GraphReviewInsight::new(
@@ -322,8 +323,8 @@ fn assemble_hooks(
(hook_registry, session_end_budget)
}
-/// The hook that reports the agent's state to the Herdr pane Luxe runs in, so its sidebar tracks
-/// Luxe. `None` outside Herdr, under `LUXE_HERDR=0`, or with `"herdr": false`.
+/// The hook that reports the agent's state to the Herdr pane luxe runs in, so its sidebar tracks
+/// luxe. `None` outside Herdr, under `LUXE_HERDR=0`, or with `"herdr": false`.
fn herdr_hook(config: &luxe_core::config::Config) -> Option {
config.herdr().then(luxe_core::hooks::herdr::HerdrHook::detect).flatten()
}
diff --git a/crates/luxe-cli/src/engine_setup/config.rs b/crates/luxe-cli/src/engine_setup/config.rs
index 080893d..48d5446 100644
--- a/crates/luxe-cli/src/engine_setup/config.rs
+++ b/crates/luxe-cli/src/engine_setup/config.rs
@@ -1,7 +1,7 @@
//! Configuration loading — auth/models, permissions, command hooks, and env resolution.
//!
//! Every reader treats its file as read-if-present: a missing or malformed config degrades to
-//! defaults rather than failing, so Luxe runs the same with or without user config.
+//! defaults rather than failing, so luxe runs the same with or without user config.
use std::path::PathBuf;
@@ -65,7 +65,7 @@ pub(crate) fn load_luxe_config() -> ModelsConfig {
/// Warn if a secrets file is readable by group or other.
///
-/// Luxe writes its own secret files `0600` (MCP token store, review stores), but `auth.json`/
+/// luxe writes its own secret files `0600` (MCP token store, review stores), but `auth.json`/
/// `models.json` are user-authored, so we can only nudge: ssh and aws-cli take the same stance
/// (warn/refuse on loose credential permissions). Unix-only; a no-op elsewhere.
#[cfg(unix)]
@@ -230,7 +230,7 @@ pub(crate) fn global_permissions_path() -> PathBuf {
luxe_config_dir().join("permissions.json")
}
-/// Luxe's own config directory. See [`luxe_core::user_dirs::luxe_config_dir`], the one owner.
+/// luxe's own config directory. See [`luxe_core::user_dirs::luxe_config_dir`], the one owner.
pub(crate) fn luxe_config_dir() -> PathBuf {
luxe_core::user_dirs::luxe_config_dir()
}
diff --git a/crates/luxe-cli/src/engine_setup/provider.rs b/crates/luxe-cli/src/engine_setup/provider.rs
index 7ecb73a..d949d8a 100644
--- a/crates/luxe-cli/src/engine_setup/provider.rs
+++ b/crates/luxe-cli/src/engine_setup/provider.rs
@@ -1,4 +1,4 @@
-//! Provider adapter construction — the active adapter plus a registry of every provider Luxe can
+//! Provider adapter construction — the active adapter plus a registry of every provider luxe can
//! route to, so an in-session model switch or a mid-session `/login` reaches the right backend.
use std::collections::HashMap;
@@ -88,7 +88,7 @@ pub(super) fn active_auth(
/// the provider's own variable, then the key its block names.
///
/// One chain for the provider a session starts on and every one it can switch to, so the key that
-/// goes out never depends on how the session got there. A provider Luxe does not know has no
+/// goes out never depends on how the session got there. A provider luxe does not know has no
/// variable of its own, only the `OPENAI_API_KEY` fallback, so its configured key comes first: an
/// OpenAI key must not go to someone else's endpoint in place of the key written for it. Keys stay
/// as written; the source expands them per request.
@@ -108,7 +108,7 @@ fn provider_auth(store: AuthStore, name: &str, config: Option<&ProviderConfig>)
/// Assemble the provider registry the engine routes through on an in-session model switch.
///
-/// Every provider Luxe knows about is registered: the built-in catalog plus anything declared in
+/// Every provider luxe knows about is registered: the built-in catalog plus anything declared in
/// `~/.config/luxe/{auth,models}.json`. That is what makes `/login` take effect mid-session — the
/// adapter is already there, and its store-backed auth source finds the new credential on the next
/// turn. A provider with no credential registers too, and answers with an actionable auth error
@@ -154,7 +154,7 @@ pub(crate) fn build_provider_registry(
/// A `--model` that names its provider, `anthropic/claude-opus-5-5`, split into the two.
///
-/// Only when the prefix is a provider Luxe knows, built in or configured, and no other provider
+/// Only when the prefix is a provider luxe knows, built in or configured, and no other provider
/// was named: under `--provider openrouter`, `anthropic/claude-x` is OpenRouter's own id, slash and
/// all. Without the split the whole string went to the default provider as an id, which is the
/// form the "no model selected" error tells you to type.
diff --git a/crates/luxe-cli/src/engine_setup/provider/tests.rs b/crates/luxe-cli/src/engine_setup/provider/tests.rs
index 43f5ea9..3638683 100644
--- a/crates/luxe-cli/src/engine_setup/provider/tests.rs
+++ b/crates/luxe-cli/src/engine_setup/provider/tests.rs
@@ -7,7 +7,7 @@ fn the_catalog_decides_which_adapter_a_known_provider_uses() {
assert_eq!(resolve_api("openai", None), ApiKind::OpenaiResponses);
assert_eq!(resolve_api("openai-codex", None), ApiKind::OpenaiCodex);
assert_eq!(resolve_api("groq", None), ApiKind::OpenaiCompletions);
- // Luxe's historical default provider name is an alias of `openai`.
+ // luxe's historical default provider name is an alias of `openai`.
assert_eq!(resolve_api("openai_compat", None), ApiKind::OpenaiResponses);
// An unknown provider is assumed to speak the most common dialect.
assert_eq!(resolve_api("some-proxy", None), ApiKind::OpenaiCompletions);
@@ -266,7 +266,7 @@ async fn the_launch_providers_command_key_runs_for_each_request() {
}
/// Launch and a `/model` switch find a provider's key by one chain: a sign-in first, then the
-/// provider's own variable, then the key its block names, and for a provider Luxe does not know,
+/// provider's own variable, then the key its block names, and for a provider luxe does not know,
/// its block before `OPENAI_API_KEY`. So a configured key always beats a key meant for another
/// provider, and which one goes out never depends on how the session reached the provider.
#[test]
@@ -329,7 +329,7 @@ fn launch_and_a_model_switch_send_the_same_key() {
);
}
-/// Which key each kind of provider settles on with nothing stored: a provider Luxe does not know
+/// Which key each kind of provider settles on with nothing stored: a provider luxe does not know
/// takes the key written for it before the `OPENAI_API_KEY` fallback, and one it knows takes its
/// own variable first.
#[test]
diff --git a/crates/luxe-cli/src/engine_setup/services.rs b/crates/luxe-cli/src/engine_setup/services.rs
index 84331b1..ecbf893 100644
--- a/crates/luxe-cli/src/engine_setup/services.rs
+++ b/crates/luxe-cli/src/engine_setup/services.rs
@@ -1,7 +1,7 @@
//! Warm, process-wide service singletons — the LSP client, code graph, DAP debugger, and
//! diagnostics cache — plus the per-project session store.
//!
-//! Luxe runs one project per process, so each service is a `OnceLock` singleton reused across
+//! luxe runs one project per process, so each service is a `OnceLock` singleton reused across
//! every tool call and hook instead of being spawned per request.
use std::sync::Arc;
@@ -56,7 +56,7 @@ pub(crate) fn tool_statuses() -> Vec {
/// Three inputs, and the judgement is here rather than in `luxe-lsp` so it can be read and tested
/// in one place: the user's preference, whether a front-end can prompt, and `--trust-project`.
///
-/// `interactive` carries the whole argument for the default. Luxe is a coding agent someone sits
+/// `interactive` carries the whole argument for the default. luxe is a coding agent someone sits
/// in front of; the loop is a feature, not the usual way it runs. With a person there, using the
/// server the project pinned is ordinary work — and `node_modules/.bin` exists only because
/// `npm install` already ran that project's install scripts, which is arbitrary code and a wider
@@ -96,7 +96,7 @@ pub(crate) fn project_tooling_allowed() -> bool {
/// The process-wide warm LSP service. A singleton (like [`shared_diagnostics_store`]) so the tools,
/// the diagnostics-on-edit hook, and the workspace collector (which reads running-server state for
-/// `/lsp`) all share one instance. Luxe runs one project per process; the collector scopes
+/// `/lsp`) all share one instance. luxe runs one project per process; the collector scopes
/// running servers to the active root.
pub(crate) fn build_lsp_service() -> SharedLspService {
static SERVICE: std::sync::OnceLock = std::sync::OnceLock::new();
@@ -112,7 +112,7 @@ pub(crate) fn build_lsp_service() -> SharedLspService {
/// The process-wide warm code graph service, shared by the `impact` / `module_report` tools and the
/// graph-invalidation hook. Singleton like the LSP/DAP services; rooted at the workspace on first
-/// use (Luxe runs one project per process).
+/// use (luxe runs one project per process).
pub(crate) fn build_graph_service(
repo_root: &std::path::Path,
) -> Arc {
@@ -223,7 +223,7 @@ fn read_config_text(path: &std::path::Path) -> Result, String> {
///
/// A shared handle at the binary boundary: it lets the (core) hook publish per-file diagnostics
/// that the (core-agnostic) collector reads for the Context Dock without threading a channel
-/// through every front-end. Luxe runs one project per process; the collector filters entries to the
+/// through every front-end. luxe runs one project per process; the collector filters entries to the
/// active root, so a project switch never leaks another tree's diagnostics.
pub(crate) fn shared_diagnostics_store() -> luxe_core::lsp::DiagnosticsStore {
static STORE: std::sync::OnceLock =
diff --git a/crates/luxe-cli/src/engine_setup/session.rs b/crates/luxe-cli/src/engine_setup/session.rs
index 6c3b769..182b616 100644
--- a/crates/luxe-cli/src/engine_setup/session.rs
+++ b/crates/luxe-cli/src/engine_setup/session.rs
@@ -169,7 +169,7 @@ pub(super) fn attach_session(
Ok((session_id, claim))
}
-/// Claim `session_id` for this process, or say which Luxe has it open and how to get there.
+/// Claim `session_id` for this process, or say which luxe has it open and how to get there.
fn claim_session(
store: &luxe_core::session::JsonlSessionStore,
session_id: &SessionId,
@@ -181,7 +181,7 @@ fn claim_session(
}
}
-/// Why `session` can't be opened here right now, when another Luxe has it open.
+/// Why `session` can't be opened here right now, when another luxe has it open.
///
/// Asked before leaving the session that's running, so the reader stays where they are rather
/// than landing outside the TUI. This process's own claims don't count against it.
@@ -195,7 +195,7 @@ pub(crate) fn open_elsewhere_now(state_root: &std::path::Path, session: &str) ->
}
}
-/// Where the session is open, when the Luxe that has it says so, and what to do about it.
+/// Where the session is open, when the luxe that has it says so, and what to do about it.
fn open_elsewhere(id: &SessionId) -> String {
let runtime = luxe_core::user_dirs::luxe_runtime_dir();
let holder = luxe_core::local_endpoint::live_sessions(&runtime)
@@ -207,7 +207,7 @@ fn open_elsewhere(id: &SessionId) -> String {
};
// `luxe join` is not offered: it speaks the protocol, so a person running it sees JSON
format!(
- "session {id} is already open in another Luxe{place}. Carry on there, or close it \
+ "session {id} is already open in another luxe{place}. Carry on there, or close it \
there first"
)
}
@@ -299,7 +299,7 @@ mod tests {
};
let refused = attach_session(&settings, &mut store).expect_err("held elsewhere");
- let expected = format!("session {id} is already open in another Luxe");
+ let expected = format!("session {id} is already open in another luxe");
assert!(refused.to_string().contains(&expected), "{refused}");
assert!(open_elsewhere_now(dir.path(), &id.0).is_some());
drop(other_process);
diff --git a/crates/luxe-cli/src/engine_setup/settings.rs b/crates/luxe-cli/src/engine_setup/settings.rs
index e5d7f38..f60175e 100644
--- a/crates/luxe-cli/src/engine_setup/settings.rs
+++ b/crates/luxe-cli/src/engine_setup/settings.rs
@@ -153,16 +153,17 @@ impl AgentSettings {
/// Resolve settings, letting `cli` overrides win over env and config.
pub(crate) fn resolve(repo_root: PathBuf, cli: CliOverrides) -> Self {
let config = load_luxe_config();
- // What the last session picked, unless the user asked us not to remember. It sits *below*
- // the CLI and the environment (a one-off `--model` must stay a one-off) and *above* the
- // configured default, which is the point: the default is where you start, not where you are
- // pinned.
+ // What the last session picked in this project, or anywhere when this project has no pick
+ // of its own, unless the user asked us not to remember. It sits *below* the CLI and the
+ // environment (a one-off `--model` must stay a one-off) and *above* the configured
+ // default, which is the point: the default is where you start, not where you are pinned.
let ui_config = luxe_core::config::Config::load(&repo_root);
let remembered = ui_config
.agent
.remember_model()
.then(luxe_core::runtime_state::RuntimeState::load)
.unwrap_or_default();
+ let last_chosen = remembered.model_for(&repo_root);
let named_provider = cli.provider.or_else(|| env_nonempty("LUXE_PROVIDER"));
let from_flag_or_env = cli.model.clone().or_else(|| env_nonempty("LUXE_MODEL"));
@@ -175,17 +176,14 @@ impl AgentSettings {
};
let provider = named_provider
.or(provider_in_model)
- .or_else(|| remembered.model.as_ref().map(|model| model.provider.clone()))
+ .or_else(|| last_chosen.map(|model| model.provider.clone()))
.or_else(|| config.default_provider.clone())
.unwrap_or_else(|| "openai_compat".to_string());
// A remembered model only applies to the provider it was chosen for: pairing last session's
// Anthropic model id with a locally forced provider would resolve to nothing.
- let remembered_model = remembered
- .model
- .as_ref()
- .filter(|model| model.provider == provider)
- .map(|model| model.id.clone());
+ let remembered_model =
+ last_chosen.filter(|model| model.provider == provider).map(|model| model.id.clone());
// A configured `default_model` belongs to the provider it was configured alongside. Pairing
// it with a provider forced on the command line produces combinations that don't exist —
// `claude-sonnet-5` on Codex — so it only applies when that provider can actually serve it.
diff --git a/crates/luxe-cli/src/engine_setup/tests.rs b/crates/luxe-cli/src/engine_setup/tests.rs
index 672527d..eb74881 100644
--- a/crates/luxe-cli/src/engine_setup/tests.rs
+++ b/crates/luxe-cli/src/engine_setup/tests.rs
@@ -131,7 +131,7 @@ fn tool_registry_has_builtins_and_lsp() {
/// The loop is opt-in, so nothing it owns may reach an ordinary session.
///
/// `loop_control` was registered unconditionally, and since it is not a core tool it landed in
-/// the "available on demand" catalog of every prompt Luxe sent — offering a plain chat session a
+/// the "available on demand" catalog of every prompt luxe sent — offering a plain chat session a
/// tool for advancing phases of an autonomous run that was not happening.
#[test]
fn loop_control_is_absent_unless_a_loop_is_driving() {
@@ -189,6 +189,7 @@ fn a_remembered_model_beats_the_configured_default_but_not_a_flag() {
provider: "anthropic".into(),
id: "claude-opus-5".into(),
}),
+ project_models: Vec::new(),
thinking: Some(luxe_core::provider::ThinkingLevel::High),
recent_commands: Vec::new(),
panes: Default::default(),
@@ -213,6 +214,40 @@ fn a_remembered_model_beats_the_configured_default_but_not_a_flag() {
assert_eq!(overridden.chosen().r#ref.id, "claude-sonnet-5");
}
+/// The model last chosen in this project outranks the one last chosen anywhere, and a project with
+/// no pick of its own starts on the latter.
+#[test]
+fn a_projects_own_pick_beats_the_last_one_anywhere() {
+ use luxe_core::runtime_state::{ProjectModel, RememberedModel, RuntimeState};
+ let _guard = ENV_LOCK.lock().unwrap_or_else(|error| error.into_inner());
+ let home = tempfile::tempdir().expect("temp home");
+ let state = home.path().join("state");
+ std::fs::create_dir_all(&state).expect("state dir");
+ let here = tempfile::tempdir().expect("temp project");
+ let model =
+ |provider: &str, id: &str| RememberedModel { provider: provider.into(), id: id.into() };
+ RuntimeState {
+ model: Some(model("anthropic", "claude-opus-5")),
+ project_models: vec![ProjectModel {
+ project: luxe_core::paths::display(here.path()),
+ model: model("openai", "gpt-5.6"),
+ }],
+ ..RuntimeState::default()
+ }
+ .save_to(&state.join("state.json"))
+ .expect("write state");
+ let _state_dir = EnvGuard::set("LUXE_STATE_DIR", state.to_str().expect("utf-8 path"));
+
+ let in_project = AgentSettings::resolve(here.path().to_path_buf(), CliOverrides::default());
+ assert_eq!(
+ (in_project.provider.as_str(), in_project.chosen().r#ref.id.as_str()),
+ ("openai", "gpt-5.6")
+ );
+ let elsewhere = tempfile::tempdir().expect("another project");
+ let fresh = AgentSettings::resolve(elsewhere.path().to_path_buf(), CliOverrides::default());
+ assert_eq!(fresh.chosen().r#ref.id, "claude-opus-5", "no pick here, so the last anywhere");
+}
+
/// A remembered model must not be pasted onto a different provider.
#[test]
fn a_remembered_model_is_ignored_when_the_provider_is_forced_elsewhere() {
@@ -226,6 +261,7 @@ fn a_remembered_model_is_ignored_when_the_provider_is_forced_elsewhere() {
provider: "anthropic".into(),
id: "claude-opus-5".into(),
}),
+ project_models: Vec::new(),
thinking: None,
recent_commands: Vec::new(),
panes: Default::default(),
@@ -359,7 +395,7 @@ impl Drop for EnvGuard {
}
}
-/// Nothing configured means nothing chosen. Luxe used to reach for whichever id the provider
+/// Nothing configured means nothing chosen. luxe used to reach for whichever id the provider
/// catalog named first, which is a literal in the binary: it goes stale as models turn over,
/// and the first thing a fresh install did with it was send it.
#[test]
@@ -541,6 +577,7 @@ fn a_models_own_thinking_level_outranks_the_remembered_one() {
luxe_core::runtime_state::RuntimeState {
schema_version: luxe_core::runtime_state::RUNTIME_STATE_SCHEMA_VERSION,
model: None,
+ project_models: Vec::new(),
thinking: Some(ThinkingLevel::XHigh),
recent_commands: Vec::new(),
panes: Default::default(),
diff --git a/crates/luxe-cli/src/engine_setup/tools.rs b/crates/luxe-cli/src/engine_setup/tools.rs
index eb4ec63..8196854 100644
--- a/crates/luxe-cli/src/engine_setup/tools.rs
+++ b/crates/luxe-cli/src/engine_setup/tools.rs
@@ -54,7 +54,7 @@ pub(crate) struct ToolDenyReport {
removed: Vec,
/// Essential, so the request was refused
refused: Vec,
- /// A tool Luxe knows, not registered on this machine
+ /// A tool luxe knows, not registered on this machine
unavailable: Vec,
/// No tool of that name anywhere, with the nearest one when there is a near miss
unknown: Vec<(DeniedTool, Option<&'static str>)>,
@@ -162,7 +162,7 @@ pub(crate) fn apply_tool_deny_list(
report
}
-/// Every tool Luxe can register, whatever this machine happens to have installed.
+/// Every tool luxe can register, whatever this machine happens to have installed.
///
/// The set that makes "no such tool" tellable from "not here": four tools register only where
/// their program exists (`debug`, `screenshot`, `ast_grep`) or where something is configured
@@ -344,7 +344,7 @@ pub(crate) async fn register_mcp_tools(registry: &mut ToolRegistry, repo_root: &
/// The MCP servers this run may start, and the ids of any the project declared that it may not.
///
/// A stdio server in `.mcp.json` is a command the repository chose, and an `eager` one must not
-/// start the moment Luxe opens a cloned tree, headless included. It answers to the gate project
+/// start the moment luxe opens a cloned tree, headless included. It answers to the gate project
/// language servers and formatters already use, not a second one.
pub(crate) fn mcp_servers_for_run(
repo_root: &std::path::Path,
diff --git a/crates/luxe-cli/src/engine_setup/tools/tests.rs b/crates/luxe-cli/src/engine_setup/tools/tests.rs
index f9466c1..fe2c719 100644
--- a/crates/luxe-cli/src/engine_setup/tools/tests.rs
+++ b/crates/luxe-cli/src/engine_setup/tools/tests.rs
@@ -108,7 +108,7 @@ fn the_deny_list_classifies_what_a_real_run_asks_for() {
assert_eq!(
report.unavailable.iter().map(|d| d.name.as_str()).collect::>(),
["mcp"],
- "a tool Luxe knows that this run never registered is not a typo"
+ "a tool luxe knows that this run never registered is not a typo"
);
assert_eq!(report.unknown.len(), 1);
assert_eq!(report.unknown[0].0.name, "websearch");
diff --git a/crates/luxe-cli/src/guide.md b/crates/luxe-cli/src/guide.md
index e0070d9..040760b 100644
--- a/crates/luxe-cli/src/guide.md
+++ b/crates/luxe-cli/src/guide.md
@@ -4,7 +4,7 @@
title: "How the agent works"
summary: "How the agent writes to you, asks you questions, plans, and gets a second opinion."
order: 11
-group: "Using Luxe"
+group: "Using luxe"
---
## How it writes
@@ -62,7 +62,7 @@ nor the tools, so it does not re-read the conversation at full price.
In plan mode the agent asks its open questions one at a time, then writes the plan as one
`## Phase N` section per phase, numbered from 0, which is the shape [`luxe loop`](loop.md) works
-through. When a turn writes the plan, Luxe asks what next:
+through. When a turn writes the plan, luxe asks what next:
- **Build it here**, with the planning conversation still in context
- **Build it in a fresh session**, starting from `PLAN.md` alone
@@ -115,10 +115,12 @@ In order, the first that is set:
1. `--model` and `--thinking` on the command line
2. `LUXE_MODEL` and `LUXE_PROVIDER` in the environment
-3. The model you last chose
-4. `default_model` in `models.json`
+3. The model you last chose in this project
+4. The model you last chose anywhere
+5. `default_model` in `models.json`
-Luxe never picks a model for you. With none of these set, sending a message opens the model picker,
+luxe never picks a model for you, and signing in doesn't either: the model picker opens at the top
+of the list for you to choose. With none of these set, sending a message opens the model picker,
and `--headless` and the loop refuse to start. To always start from your configured default rather
than your last choice:
@@ -153,13 +155,13 @@ Every setting on this page goes in `~/.config/luxe/config.json`, or a project's
`/theme` previews each theme as you move through the list. `enter` keeps one and saves it to your
config; `esc` puts back the theme you had.
-Luxe's own theme is `luxe`, a warm graphite with a brass accent. Every accent clears a 4.5:1
+luxe's own theme is `luxe`, a warm graphite with a brass accent. Every accent clears a 4.5:1
contrast ratio against the background, and body text 7:1. The others are dark versions of
community favorites and of Omarchy's themes:
| | |
| --- | --- |
-| Luxe | `luxe` |
+| luxe | `luxe` |
| Editors | `jetbrains` (Darcula), `vscode` (Visual Studio 2017 Dark), `xcode` (Default Dark) |
| Community | `aura`, `ayu`, `catppuccin`, `dracula`, `everforest`, `github` (Dark Dimmed), `gruvbox`, `kanagawa`, `monokai`, `noctis-bordo`, `nord`, `one-dark`, `panda`, `rose-pine`, `tokyo-night`, `vesper`, `zenburn` |
| Omarchy | `ethereal`, `hackerman`, `last-horizon`, `lumon`, `matte-black`, `miasma`, `osaka-jade`, `retro-82`, `ristretto`, `solitude`, `vantablack` |
@@ -168,7 +170,7 @@ community favorites and of Omarchy's themes:
{ "theme": "nord" }
```
-An unknown name falls back to `luxe`, and Luxe says so when the session opens. See them all on
+An unknown name falls back to `luxe`, and luxe says so when the session opens. See them all on
[luxe.dev](https://luxe.dev/#craft).
Code is colored by the theme too: in answers, tool cards, diffs and the preview, keywords take the
@@ -196,7 +198,7 @@ The names are the ones a theme file uses; [`luxe.json`](https://github.com/evanw
lists every one.
One more name is optional: `modePlan`, the color [plan mode](agent.md#plan-then-build) wears on the
-composer and the menus and questions around it. Without it, Luxe picks the first of the theme's
+composer and the menus and questions around it. Without it, luxe picks the first of the theme's
other colors that reads as different from its accent.
### Make a theme
@@ -218,12 +220,12 @@ for a complete one.
### On Omarchy
-Leave `theme` unset, or set it to `omarchy`, and Luxe wears your desktop's theme and changes with
+Leave `theme` unset, or set it to `omarchy`, and luxe wears your desktop's theme and changes with
it, within a couple of seconds, in a running session. It uses, in order:
1. A `luxe.json` in the current Omarchy theme. To give every theme one, write a template at
`~/.config/omarchy/themed/luxe.json.tpl` and Omarchy renders it from each theme's colors.
-2. Luxe's built-in theme of the same name, when both are dark.
+2. luxe's built-in theme of the same name, when both are dark.
3. A theme made on the spot from the Omarchy theme's `colors.toml`, with text lifted until it is
readable. This is how a light Omarchy theme comes out.
@@ -279,12 +281,12 @@ When the bar runs out of room, it drops modules in this order: `cwd`, `tokens`,
## Icons and fonts
-`glyphs` sets which icons Luxe draws. Every icon has a word beside it, so nothing is lost at any
+`glyphs` sets which icons luxe draws. Every icon has a word beside it, so nothing is lost at any
setting.
| `glyphs` | Draws | Needs |
| --- | --- | --- |
-| `auto` | `nerd` when Luxe can confirm your font has every icon, otherwise `unicode` | |
+| `auto` | `nerd` when luxe can confirm your font has every icon, otherwise `unicode` | |
| `nerd` | Nerd Font icons | A Nerd Font |
| `nerd_wide` | The same icons, each given two cells | A Nerd Font whose icons are drawn wide |
| `unicode` | Plain shapes: `■ ► ▪ ○ ●` | UTF-8 |
@@ -299,7 +301,7 @@ which brings its own. On macOS and Windows without fontconfig, it picks `unicode
have a Nerd Font. If icons sit flush against the word beside them, you have the wide build of your Nerd
Font: switch to its `Mono` build (see [Fonts](install.md#fonts)), or set `nerd_wide`.
-Fonts that draw every character Luxe uses, measured against the font files:
+Fonts that draw every character luxe uses, measured against the font files:
| Font | Missing | Width |
| --- | --- | --- |
@@ -375,7 +377,7 @@ Screenshots and images the agent reads are drawn in the conversation where the t
```
Double-click an image, or select its message (`alt+k`) and press `alt+i`, to view it full-window.
-`/images` says which setting is in effect and why. A change to `images` applies the next time Luxe
+`/images` says which setting is in effect and why. A change to `images` applies the next time luxe
starts.
## The start screen
@@ -454,7 +456,7 @@ luxe --headless -p "list the TODO comments in src/" --format text
```
A headless run needs `-p` and a model: `--model`, `LUXE_MODEL`, your last choice or `default_model`
-in `models.json`. It refuses to start without either. It exits with status 1 when the turn fails or Luxe stops before the turn ends, so a
+in `models.json`. It refuses to start without either. It exits with status 1 when the turn fails or luxe stops before the turn ends, so a
script can tell a failure from an answer without reading the output.
## Commands
@@ -466,15 +468,15 @@ script can tell a failure from an answer without reading the output.
| `luxe auth status [provider]` | A credential's type, where it came from, and when it expires |
| `luxe auth logout ` | Remove a stored credential |
| `luxe models list` | Models you can use. `--provider ` filters, `--configured` lists only the models `models.json` or `auth.json` names. |
-| `luxe models show ` | What Luxe knows about a model, and where each value came from. A bare ID uses the active provider. |
+| `luxe models show ` | What luxe knows about a model, and where each value came from. A bare ID uses the active provider. |
| `luxe models refresh` | Refresh the model catalog. `--force` even if it is fresh. |
| `luxe sessions list` | Sessions in this project, newest first |
-| `luxe join [id]` | Join a session another Luxe is running, speaking the [RPC protocol](rpc.md#join-a-running-session) on stdin and stdout. The id can be its start, or the Luxe's process id; without one, it joins the session running in this folder. |
+| `luxe join [id]` | Join a session another luxe is running, speaking the [RPC protocol](rpc.md#join-a-running-session) on stdin and stdout. The id can be its start, or the luxe's process id; without one, it joins the session running in this folder. |
| `luxe join --list` | Sessions running on this machine that can be joined |
| `luxe remote` | Serve this machine's sessions to your own phone or tablet. Listens on this machine only, for Tailscale Serve to carry to your devices, and lets in this machine's tailnet owner from a paired device. |
| `luxe remote --port ` | The port it listens on (default 4519) |
| `luxe remote --allow ` | Also let this tailnet login in, as Tailscale names it (`alex@github`). Repeatable. |
-| `luxe remote setup` | Check Tailscale, offer to publish the bridge on your tailnet with Tailscale Serve, offer to run it at login as a user service, then show a QR code to pair a device. Each step that changes something outside Luxe asks first. |
+| `luxe remote setup` | Check Tailscale, offer to publish the bridge on your tailnet with Tailscale Serve, offer to run it at login as a user service, then show a QR code to pair a device. Each step that changes something outside luxe asks first. |
| `luxe remote pair` | Show a QR code and link that pair one phone or tablet. They work once, within ten minutes. |
| `luxe remote devices` | The paired devices |
| `luxe remote devices revoke ` | Unpair a device. Its open sessions close within a few seconds. |
@@ -491,7 +493,7 @@ script can tell a failure from an answer without reading the output.
| `luxe loop fresh` | Reset the run's turn count; see [Autonomous runs](loop.md#start-a-run) |
| `luxe docs [page]` | Read this guide, offline and for the installed version. `--search ` finds lines across every page. |
| `luxe doctor` | Check which tools, language servers and debug adapters work |
-| `luxe update` | Update Luxe with the command for however it was installed; see [Updates](configuration.md#updates). `--check` only reports. |
+| `luxe update` | Update luxe with the command for however it was installed; see [Updates](configuration.md#updates). `--check` only reports. |
| `luxe init` | Set up a theme, a provider, the tools and the start screen, and write them to `~/.config/luxe/config.json`. `--defaults` skips the questions, `--force` starts from an empty file; see [Editing](configuration.md#editing). |
| `luxe wire ` | Summarize a `LUXE_WIRE_LOG` capture; see [Troubleshooting](troubleshooting.md#see-what-is-sent-to-the-model). `--tools` breaks the tool schemas' size down per tool. |
@@ -501,7 +503,7 @@ Every command also takes the model and permission flags above.
title: "Commands and keys"
summary: "The Command Palette, slash commands, Help, every default key, and how to rebind them."
order: 4
-group: "Using Luxe"
+group: "Using luxe"
---
## Command Palette and slash commands
@@ -525,7 +527,7 @@ list says what the command does and the arguments it takes.
Past the command, the list shows the arguments it takes, such as `/diff turn` or
`/diff pr `, and narrows as you type. `tab` fills in the one under the cursor.
`enter` runs it when nothing is missing, and fills it in when it still needs a value. `esc` puts the
-list away for that command. In the composer, a command Luxe knows is highlighted, and so are the
+list away for that command. In the composer, a command luxe knows is highlighted, and so are the
words of it you have typed, like `pr` in `/diff pr 142`. A command that stays plain is misspelled.
## Slash commands
@@ -565,8 +567,8 @@ words of it you have typed, like `pr` in `/diff pr 142`. A command that stays pl
| `/login`, `/logout` | Add or remove a provider's credential |
| `/auth` | Which providers you are signed in to, and how |
| `/reload` | Re-read your config and keys without restarting |
-| `/quit` | Leave Luxe. The session is already saved. |
-| `/update` | Leave Luxe and update it, as `luxe update` does |
+| `/quit` | Leave luxe. The session is already saved. |
+| `/update` | Leave luxe and update it, as `luxe update` does |
The Command Palette also lists commands with no slash form, such as **Copy message**, **Swap the
sidebar's side** and, in a review, **Send comments**.
@@ -699,7 +701,7 @@ Put your bindings in `~/.config/luxe/keybindings.json`, or open it from **Config
## Alt keys on a Mac
A Mac has no Alt key. Its Option key types characters instead: `option+o` is `ø`, `option+p` is
-`π`. Most of Luxe's `alt` keys need your terminal to send Option as Alt:
+`π`. Most of luxe's `alt` keys need your terminal to send Option as Alt:
| Terminal | Setting |
| --- | --- |
@@ -711,19 +713,19 @@ A Mac has no Alt key. Its Option key types characters instead: `option+o` is `ø
| Alacritty | `option_as_alt = "Both"` under `[window]` |
Ghostty already does this on the "U.S." keyboard layouts. On the others it sends the character and
-marks it as an Alt key. On "ABC", the Mac's default English layout, Luxe reads that back as the key
+marks it as an Alt key. On "ABC", the Mac's default English layout, luxe reads that back as the key
you pressed, so `option+o` opens a file from the preview without any setting. On other layouts, set
the option above.
@@@ page: configuration
---
title: "Configuration"
-summary: "Where Luxe's files live, what a project can set, and every key in config.json."
+summary: "Where luxe's files live, what a project can set, and every key in config.json."
order: 1
group: "Customize"
---
-Luxe works with no configuration. Everything here is optional, and most of it can be set from inside
-Luxe: **Configuration** (on the start screen, or `/config`) signs you in and opens these files in
+luxe works with no configuration. Everything here is optional, and most of it can be set from inside
+luxe: **Configuration** (on the start screen, or `/config`) signs you in and opens these files in
your editor, [`/settings`](#settings) changes the settings you use most, `/theme` sets the theme,
and `/model` the model.
@@ -752,7 +754,7 @@ in `.luxe/` at its root.
### What a project cannot change
A project's `.luxe/config.json` overrides yours key by key, with a few exceptions. These run
-programs, reach the network, or decide what a repository is trusted with, so Luxe reads them from
+programs, reach the network, or decide what a repository is trusted with, so luxe reads them from
your own config only:
- `launchers` and `mentions`, which run programs
@@ -761,7 +763,7 @@ your own config only:
- `agent.on_ask`, which a project can set to `deny` but not to `allow`
- `agent.cache_warming`, which a project can turn down but not up, since it spends your money
-Project `gate.commands` also run programs. Luxe ignores them and warns unless you start with
+Project `gate.commands` also run programs. luxe ignores them and warns unless you start with
`--trust-project`. With trust, a nonempty project list replaces your global list. Otherwise, the
loop uses your global commands. Move the list to your own config if you don't want to trust the
project.
@@ -772,8 +774,8 @@ project.
screen) and writes the answers to `config.json`, changing only those keys. It also writes a
`config.schema.json` beside it so your editor can complete and check every key. `--defaults` writes
a file that chooses nothing, without asking, and so does a run with no terminal; with a
-`config.json` already there, either one needs `--force`, which removes the old file first. Luxe ignores a key it does
-not know, so an old config always loads on a newer Luxe; check the spelling if a setting seems to
+`config.json` already there, either one needs `--force`, which removes the old file first. luxe ignores a key it does
+not know, so an old config always loads on a newer luxe; check the spelling if a setting seems to
do nothing.
`/reload` (`ctrl+shift+r`) applies changes to `config.json` and `keybindings.json` without
@@ -799,7 +801,7 @@ change writes one key to your `config.json` and leaves the rest of the file as i
Each value says where it comes from: `default`, `global` (your `config.json`), or `project` (this
project's `.luxe/config.json`, which wins over yours, so it is changed there). Most settings take
-effect at once; the rest say they take effect the next time Luxe starts. That includes the Tools
+effect at once; the rest say they take effect the next time luxe starts. That includes the Tools
page, because the tools the agent is offered are fixed for a run. In a window too small for the
panel, it says so instead, and only `esc` works until the window fits.
@@ -832,22 +834,22 @@ are doing.
The notification goes through your terminal, as an escape sequence it turns into a desktop
notification, so it works over ssh and needs nothing installed. Ghostty, iTerm2, kitty, WezTerm,
-Warp and foot show them; Luxe recognizes those and sends nothing to a terminal it does not know. If
-yours shows them and Luxe does not recognize it, name the sequence: `"notification_method": "osc9"`
+Warp and foot show them; luxe recognizes those and sends nothing to a terminal it does not know. If
+yours shows them and luxe does not recognize it, name the sequence: `"notification_method": "osc9"`
(the common one) or `"osc777"`. Inside tmux, add `set -g allow-passthrough on` to your tmux
configuration. Your terminal may also have to be allowed to send notifications in your system's
settings.
### herdr
-Inside a [herdr](https://herdr.dev) pane, Luxe tells herdr whether it is working, idle, or waiting
+Inside a [herdr](https://herdr.dev) pane, luxe tells herdr whether it is working, idle, or waiting
on you, and why: `Approve bash`, or the question it asked. herdr shows that in its sidebar, raises
-its own notification when Luxe finishes or needs you (set in herdr's `[ui.toast]`), and reports it
+its own notification when luxe finishes or needs you (set in herdr's `[ui.toast]`), and reports it
to `herdr agent wait`. After a herdr restart, the pane reopens the same session with
-`luxe --resume`. Luxe sends no escape-sequence notifications inside herdr, because herdr keeps
+`luxe --resume`. luxe sends no escape-sequence notifications inside herdr, because herdr keeps
them for itself.
-[Recipes](recipes.md#luxe-colors-in-herdr) has a Luxe palette for herdr's theme.
+[Recipes](recipes.md#luxe-colors-in-herdr) has a luxe palette for herdr's theme.
This is local to your machine and on by default. Outside herdr nothing is sent.
`LUXE_HERDR=0` in the environment stops the reports, and so does the config:
@@ -858,7 +860,7 @@ This is local to your machine and on by default. Outside herdr nothing is sent.
### Joining a session
-A running Luxe lets another program on this machine join its session: `luxe join`, a script, or
+A running luxe lets another program on this machine join its session: `luxe join`, a script, or
whatever you build on [the RPC protocol](rpc.md#join-a-running-session). Only your user account can
connect, and nothing listens on the network. To turn it off:
@@ -871,11 +873,11 @@ it back on once yours has.
### Updates
-Once a day, Luxe checks whether a newer release exists, and when one does the start screen says
+Once a day, luxe checks whether a newer release exists, and when one does the start screen says
*Update available · /update*. The check sends nothing about you or your machine; the comparison
happens locally.
-`/update`, or `luxe update` in a shell, updates Luxe with the command for however you installed
+`/update`, or `luxe update` in a shell, updates luxe with the command for however you installed
it, showing each one before it runs:
| Installed with | `luxe update` runs |
@@ -884,9 +886,10 @@ it, showing each one before it runs:
| Homebrew | `brew update`, then `brew upgrade luxe` |
| Cargo | `cargo install --locked --force luxe-cli` |
-Luxe never downloads or replaces its own program; whatever installed it does. A copy your system's
+luxe never downloads or replaces its own program; whatever installed it does. A copy your system's
package manager owns is left to it, and a downloaded archive points you at the releases page. On
-Windows, Luxe can't be replaced while it runs, so it prints the command to run once it has exited.
+Windows, a running program can't be overwritten, so luxe first renames itself to `luxe.exe.old`,
+and the next launch deletes it.
Afterwards it checks that the `luxe` your shell finds is the new version, and says so if another
copy earlier on your `PATH` is in the way. `luxe update --check` only reports.
@@ -912,9 +915,9 @@ or lazygit:
}
```
-Luxe keeps running underneath and redraws when the program exits. Text the program writes to the
+luxe keeps running underneath and redraws when the program exits. Text the program writes to the
file named by `$LUXE_HANDOFF` is put in the composer, so a gh-dash command can send
-`#123 look at this` to Luxe. A launcher's key wins over any other binding.
+`#123 look at this` to luxe. A launcher's key wins over any other binding.
## Reference
@@ -995,13 +998,13 @@ order: 6
group: "Extend"
---
-Luxe knows how to reach every provider in [Models and providers](providers.md), and the
+luxe knows how to reach every provider in [Models and providers](providers.md), and the
[models.dev](https://models.dev) catalog tells it each model's context window, output limit,
thinking levels and price. This page is for what neither covers: your own endpoint, a model the
catalog does not know, or a local server that needs its requests shaped a particular way.
Everything here goes in `~/.config/luxe/models.json`. `luxe models show /` shows what
-Luxe resolved for a model, and where each value came from.
+luxe resolved for a model, and where each value came from.
## Add an endpoint
@@ -1028,7 +1031,7 @@ Then `luxe --model gateway/qwen3-coder-480b`, or choose it in `/model`.
| Provider key | Does |
| --- | --- |
| `base_url` | The endpoint, including its version, such as `/v1` |
-| `api_key` | The key: the value, `$VAR`, `${VAR}` or `$env:VAR`, or `!command`, which runs each time a request needs the key. An unset `${VAR}` is empty, as in `mcp.json`, and Luxe logs which one. |
+| `api_key` | The key: the value, `$VAR`, `${VAR}` or `$env:VAR`, or `!command`, which runs each time a request needs the key. An unset `${VAR}` is empty, as in `mcp.json`, and luxe logs which one. |
| `api` | The API it speaks: `openai-completions` (the default), `openai-responses`, `anthropic-messages` |
| `auth` | How the key is sent: `bearer`, `x-api-key`, `api-key`, or `foundry` for both of the last two |
| `models_dev` | The models.dev provider to take model details from, when the names differ |
@@ -1036,12 +1039,12 @@ Then `luxe --model gateway/qwen3-coder-480b`, or choose it in `/model`.
| `sampling_params` | Samplers for every model it serves; see [Samplers](#samplers) |
A provider in `models.json` replaces the same provider's block in `auth.json` whole, so keep a
-provider's key and endpoint in one file. A provider Luxe already knows keeps everything you do not
-write. A key in a provider Luxe knows is used when its own variable, such as `OPENAI_API_KEY`, is
+provider's key and endpoint in one file. A provider luxe already knows keeps everything you do not
+write. A key in a provider luxe knows is used when its own variable, such as `OPENAI_API_KEY`, is
unset; for one it does not know, the key you write comes first and `OPENAI_API_KEY` is the
fallback. A key you signed in with comes before either.
-A file Luxe cannot parse is skipped with a warning in the log, and a misspelled key is ignored, so
+A file luxe cannot parse is skipped with a warning in the log, and a misspelled key is ignored, so
check the result with `luxe models show /` after editing.
## Model settings
@@ -1054,7 +1057,7 @@ each replace the catalog's whole: a `cost` you write drops the catalog's long-pr
| Key | Does |
| --- | --- |
| `context_window` | The model's context window, in tokens |
-| `max_tokens` | The most it may write in one reply. Defaults to the model's own limit from the catalog. Luxe lowers it per request to what the window has left, so setting the real limit is safe; setting a lower one is how replies get cut off. |
+| `max_tokens` | The most it may write in one reply. Defaults to the model's own limit from the catalog. luxe lowers it per request to what the window has left, so setting the real limit is safe; setting a lower one is how replies get cut off. |
| `name` | The name shown in the picker and statusline |
| `reasoning` | Whether it can think |
| `thinking` | The level a session starts at for this model |
@@ -1064,14 +1067,14 @@ each replace the catalog's whole: a `cost` you write drops the catalog's long-pr
| `compat`, `sampling_params` | Override the provider's, key by key |
A model the catalog does not know gets a 128k context window and no output limit until you set
-them. Anthropic endpoints require a limit, so there Luxe asks for 128k. If the model allows less,
-Anthropic's refusal names its real limit, and Luxe retries with that and uses it for the rest of the
+them. Anthropic endpoints require a limit, so there luxe asks for 128k. If the model allows less,
+Anthropic's refusal names its real limit, and luxe retries with that and uses it for the rest of the
session.
## Thinking levels
-`thinking_level_map` says what each of Luxe's levels sends to the model. `null` marks a level the
-model cannot do, removing it from `/thinking`; a level Luxe needs falls back to the nearest one that
+`thinking_level_map` says what each of luxe's levels sends to the model. `null` marks a level the
+model cannot do, removing it from `/thinking`; a level luxe needs falls back to the nearest one that
is left. `ultra` is the name for a provider's `max`.
```json title="~/.config/luxe/models.json"
@@ -1097,12 +1100,12 @@ is left. `ultra` is the name for a provider's `max`.
A level you do not map sends its own name, and `off` sends `none`; `xhigh` and `ultra` are offered
only once mapped. Setting a map also marks the model as one that reasons.
-An endpoint Luxe does not know sends no thinking setting at all until its `compat` says how: for
+An endpoint luxe does not know sends no thinking setting at all until its `compat` says how: for
most OpenAI-compatible servers, `"thinking_format": "openai"`, which sends `reasoning_effort` on
every request, `off` included. See [Request format](#request-format).
-For Anthropic models, Luxe uses adaptive thinking with an effort level where the model supports
-it, and a token budget for older ones. If a model refuses how thinking was requested, Luxe tries
+For Anthropic models, luxe uses adaptive thinking with an effort level where the model supports
+it, and a token budget for older ones. If a model refuses how thinking was requested, luxe tries
the other way once and remembers it for the session. `compat.force_adaptive_thinking` forces
adaptive thinking for an endpoint the catalog does not know.
@@ -1140,7 +1143,7 @@ to every model the provider serves; on a model, only the keys you write override
| `force_adaptive_thinking` | Use Anthropic's adaptive thinking |
| `supports_long_cache_retention` | Ask Anthropic to keep cached prompts for an hour rather than five minutes. Each cache write costs 2x input instead of 1.25x. Off by default, since not every Anthropic-compatible endpoint accepts it. |
| `stream_tool_arguments` | Stream a `write` call's file from Anthropic as it is written, so its card fills in live. On by default; turn off for an endpoint that rejects a tool's `eager_input_streaming` field. |
-| `defer_tool_loading` | Load a tool the agent asks for with `activate_tools` where it asked, so the cached prompt stays cached. On by default. Luxe stops on its own for an endpoint that refuses it; turn it off for one that accepts it and then misbehaves. Anthropic and Responses endpoints only. |
+| `defer_tool_loading` | Load a tool the agent asks for with `activate_tools` where it asked, so the cached prompt stays cached. On by default. luxe stops on its own for an endpoint that refuses it; turn it off for one that accepts it and then misbehaves. Anthropic and Responses endpoints only. |
| `copilot_headers`, `openrouter_headers` | Send those services' identifying headers. With `copilot_headers`, each model also goes to the Copilot endpoint it takes. |
| `use_responses_api` | Use OpenAI's Responses API. Provider only; prefer `"api": "openai-responses"`. |
@@ -1154,7 +1157,7 @@ provider's.
A local model's card often gives a sampling recipe with settings the OpenAI format has no field for,
such as `top_k`, `min_p` and `repeat_penalty`. `sampling_params` is added to the request as written,
-after Luxe's own fields, so its values win:
+after luxe's own fields, so its values win:
```json title="~/.config/luxe/models.json"
{
@@ -1202,9 +1205,9 @@ sent as written. LM Studio ignores `chat_template_kwargs`; use `"thinking_format
## Model details
-Luxe takes a model's details from three places, each overriding the one before, value by value:
+luxe takes a model's details from three places, each overriding the one before, value by value:
-1. The models.dev catalog built into Luxe, which works offline
+1. The models.dev catalog built into luxe, which works offline
2. A fresher copy, refreshed in the background into `~/.cache/luxe/models-dev.json`
3. Your `models.json`
@@ -1221,7 +1224,7 @@ anthropic/claude-opus-5
cost /Mtok $5.00/$25.00 [models.dev]
```
-Each value names where it came from: `snapshot` for the catalog built into Luxe, `models.dev` for
+Each value names where it came from: `snapshot` for the catalog built into luxe, `models.dev` for
the refreshed copy, `config` for your `models.json`, and `default` where none of them said.
| Variable | Does |
@@ -1235,7 +1238,7 @@ the refreshed copy, `config` for your `models.json`, and `default` where none of
### Keyless sign-in with Microsoft Entra ID
-If your Foundry resource has key sign-in turned off, give Luxe a command that prints a token:
+If your Foundry resource has key sign-in turned off, give luxe a command that prints a token:
```json title="~/.config/luxe/models.json"
{
@@ -1255,9 +1258,9 @@ Run `az login` first. Only the Claude provider needs `auth`; the others already
token. The command runs for each request and nothing is kept, so a token from `az`, which lasts
about an hour, is replaced as it expires and a long session carries on.
-### Choose which deployments Luxe offers
+### Choose which deployments luxe offers
-Luxe records what a resource has deployed in each provider's `served_models`, and the model behind
+luxe records what a resource has deployed in each provider's `served_models`, and the model behind
any deployment named something else in `deployments`, which is where it looks up the context window
and price. If your key cannot list deployments, or you want to choose the list yourself, turn
discovery off and write both:
@@ -1274,7 +1277,7 @@ discovery off and write both:
}
```
-`discover_deployments: false` on any of the three providers stops Luxe asking the resource at all.
+`discover_deployments: false` on any of the three providers stops luxe asking the resource at all.
### Thinking on partner models
@@ -1296,21 +1299,21 @@ Use `reasoning_effort` only: Foundry refuses DeepSeek's own `thinking` format. F
report these models' thinking tokens, so `/cost` counts them as output.
A Foundry deployment can have a different context window and price from the vendor's own API.
-Luxe uses Azure's figures for Foundry models.
+luxe uses Azure's figures for Foundry models.
## OpenRouter
-- Luxe asks OpenRouter to cache prompts for Claude models, which it otherwise does not, and which
+- luxe asks OpenRouter to cache prompts for Claude models, which it otherwise does not, and which
makes later turns of a session far cheaper. `"cache_control_format": null` in the provider's
`compat` stops it.
-- Some models on OpenRouter cannot turn thinking off. For those, Luxe leaves the setting out and
+- Some models on OpenRouter cannot turn thinking off. For those, luxe leaves the setting out and
the model thinks at its own default. `luxe models show openrouter/` says `never off`.
-- Luxe identifies itself to OpenRouter with its `HTTP-Referer` and `X-Title` headers, which carry
+- luxe identifies itself to OpenRouter with its `HTTP-Referer` and `X-Title` headers, which carry
nothing about you. `"openrouter_headers": false` in `compat` drops them.
@@@ page: environment
---
title: "Environment variables"
-summary: "The environment variables you can set, and the ones Luxe sets."
+summary: "The environment variables you can set, and the ones luxe sets."
order: 3
group: "Reference"
---
@@ -1338,10 +1341,10 @@ config file turned off.
| Variable | Does |
| --- | --- |
| `LUXE_CONFIG_DIR` | Where configuration and credentials live, instead of `~/.config/luxe` |
-| `LUXE_STATE_DIR` | Where Luxe keeps its state, such as the model you last chose, MCP sign-ins and the logs of commands that only report, instead of `~/.local/state/luxe` |
+| `LUXE_STATE_DIR` | Where luxe keeps its state, such as the model you last chose, MCP sign-ins and the logs of commands that only report, instead of `~/.local/state/luxe` |
| `XDG_CONFIG_HOME`, `XDG_STATE_HOME`, `XDG_CACHE_HOME` | Honored on every platform; see [Install](install.md#where-luxe-keeps-files) |
-When Luxe works out where its files go, a path in one of these, or in `HOME`, that is not valid
+When luxe works out where its files go, a path in one of these, or in `HOME`, that is not valid
UTF-8 is ignored, as if the variable were unset.
## Model catalog
@@ -1368,7 +1371,7 @@ UTF-8 is ignored, as if the variable were unset.
| `LUXE_LAZY_TOOLS` | `1` | `0` sends every tool's description with every request |
| `LUXE_CORE_TOOLS` | | The tools sent with every request, comma-separated; the rest load when needed |
| `LUXE_DAP_STOP_TIMEOUT_SECS` | `30` | How long the debugger waits for a program to stop |
-| `LUXE_MAX_DEPTH` | `2` | How deeply Luxe may run inside itself, as with [subagents](recipes.md#subagents) |
+| `LUXE_MAX_DEPTH` | `2` | How deeply luxe may run inside itself, as with [subagents](recipes.md#subagents) |
| `LUXE_WEB_ALLOW_PRIVATE` | | `1` lets `web_fetch` and `screenshot` reach your local network and files |
| `LUXE_WEB_TIMEOUT_SECS` | `30` | How long one fetch may take. Past 60, raise `LUXE_TOOL_TIMEOUT_SECS` too. |
| `BRAVE_API_KEY`, `TAVILY_API_KEY`, `EXA_API_KEY`, `FIRECRAWL_API_KEY` | | Search providers, and Firecrawl for pages built with JavaScript; see [Tools](tools.md#web) |
@@ -1377,12 +1380,12 @@ UTF-8 is ignored, as if the variable were unset.
| Variable | Does |
| --- | --- |
-| `LUXE_EDITOR`, then `VISUAL`, then `EDITOR` | The editor Luxe opens files in |
+| `LUXE_EDITOR`, then `VISUAL`, then `EDITOR` | The editor luxe opens files in |
| `LUXE_FILE_MANAGER`, then `FILE_MANAGER` | The file manager `O` in the dock's file tree opens a folder in. Without either, the first of yazi, lf, ranger and nnn that is installed, then your desktop's own. |
| `LUXE_HERDR` | `0` stops reporting state to [Herdr](recipes.md#run-luxe-in-herdr) |
| `LUXE_UPDATE_CHECK` | `0` turns off the update check |
| `LUXE_JOINABLE` | `0` stops other programs [joining](configuration.md#joining-a-session) this run's session |
-| `LUXE_RUNTIME_DIR` | Where a running Luxe keeps the socket and record that let a program join it, instead of `luxe` in `$XDG_RUNTIME_DIR` on Linux or `$TMPDIR` on macOS, or `run` in the state folder |
+| `LUXE_RUNTIME_DIR` | Where a running luxe keeps the socket and record that let a program join it, instead of `luxe` in `$XDG_RUNTIME_DIR` on Linux or `$TMPDIR` on macOS, or `run` in the state folder |
## Diagnostics
@@ -1391,22 +1394,22 @@ UTF-8 is ignored, as if the variable were unset.
| `LUXE_LOG` | What goes in the log, such as `debug` or `luxe_core=debug`. `RUST_LOG` works too. See [Logs](troubleshooting.md#logs). |
| `LUXE_WIRE_LOG` | A file to record every request sent to the model; see [Troubleshooting](troubleshooting.md#see-what-is-sent-to-the-model) |
-## Set by Luxe
+## Set by luxe
| Variable | Is |
| --- | --- |
-| `LUXE_DEPTH` | How deeply this run is nested inside another Luxe |
-| `LUXE_REVIEW_GATE_ACTIVE` | Set for a loop's [gate commands](loop.md#check-each-phase), so a Luxe they start does not run a gate of its own |
+| `LUXE_DEPTH` | How deeply this run is nested inside another luxe |
+| `LUXE_REVIEW_GATE_ACTIVE` | Set for a loop's [gate commands](loop.md#check-each-phase), so a luxe they start does not run a gate of its own |
| `LUXE_HANDOFF` | For a [launcher](configuration.md#launchers): a file whose contents go in the composer when the program exits |
@@@ page: extending
---
-title: "Extending Luxe"
+title: "Extending luxe"
summary: "Which kind of extension you need, how to install one, and what stays stable across updates."
order: 1
group: "Extend"
---
-Luxe extends through files and programs, never through code loaded into Luxe itself. A file carries
+luxe extends through files and programs, never through code loaded into luxe itself. A file carries
instructions or settings; a program carries behavior, and runs with the same access you have.
Installing an extension is copying a file.
@@ -1414,13 +1417,13 @@ Installing an extension is copying a file.
| To | Use | Runs |
| --- | --- | --- |
-| Change how Luxe looks, or which tools exist | [Configuration](configuration.md) and [themes](appearance.md#make-a-theme) | Nothing |
+| Change how luxe looks, or which tools exist | [Configuration](configuration.md) and [themes](appearance.md#make-a-theme) | Nothing |
| Teach the agent a procedure or house style | [Skills](skills.md) and [`AGENTS.md`](project-context.md) | Nothing |
| Point the agent at your tracker or runbooks with `@` | [Mention sources](mention-sources.md) | Your program, when its tab opens |
| Give the agent new tools | [An MCP server](mcp.md) | Your program, when the agent calls it |
| Block, rewrite, audit or react to what the agent does | [Hooks](hooks.md) | Your program, at points in a turn |
-| Use an endpoint or model Luxe does not know | [Custom providers](custom-providers.md) | Nothing, or your `!command` for a key |
-| Drive Luxe from your own program, or build a new front end: conversation, Guided Review and GitHub | [The RPC protocol](rpc.md) | Your program, driving Luxe |
+| Use an endpoint or model luxe does not know | [Custom providers](custom-providers.md) | Nothing, or your `!command` for a key |
+| Drive luxe from your own program, or build a new front end: conversation, Guided Review and GitHub | [The RPC protocol](rpc.md) | Your program, driving luxe |
| Open another terminal tool with a key | [Launchers](configuration.md#launchers) | Your program, on a key |
The rows are in order of effort; the first that fits is usually right. [Recipes](recipes.md) puts
@@ -1465,7 +1468,7 @@ releases.
`.luxe/` writes its own `.gitignore`, which keeps everything in it out of Git except
`permissions.json`, `mcp.json`, `hooks.json` and `skills/`. To share more with your team, such as the
-project's `config.json`, write that `.gitignore` yourself: Luxe rewrites its own copy, but leaves
+project's `config.json`, write that `.gitignore` yourself: luxe rewrites its own copy, but leaves
alone one it did not write.
## Trust
@@ -1474,11 +1477,11 @@ alone one it did not write.
`gate.commands` run only with `--trust-project`, and its permission rules can only make yours
stricter. MCP servers, language servers and formatters it brings start, by default, only in an
interactive session or with `--trust-project`; `lsp.project_servers` in your own config changes
-that, see [Tools](tools.md#a-server-the-project-installed). This is about which programs Luxe starts. A
+that, see [Tools](tools.md#a-server-the-project-installed). This is about which programs luxe starts. A
language server itself may run project code, as rust-analyzer does with `build.rs`, so open an
-unfamiliar repository in Luxe the way you would build it.
+unfamiliar repository in luxe the way you would build it.
-**A repository can still write instructions.** Its `AGENTS.md` and skills reach the model. Luxe
+**A repository can still write instructions.** Its `AGENTS.md` and skills reach the model. luxe
removes text a reviewer cannot see, but plain text is still instructions, so read an unfamiliar
repository's `AGENTS.md` as you would its `Makefile`.
@@ -1487,7 +1490,7 @@ repository's `AGENTS.md` as you would its `Makefile`.
| | Across updates |
| --- | --- |
| `config.json`, `models.json`, `keybindings.json`, `lsp-servers.json`, `debug-adapters.json` | **Only added to.** An old file always loads. An unknown key is ignored, and no key changes meaning. |
-| Sessions in `.luxe/sessions/` | A newer Luxe reads every older session. An older Luxe skips what it does not know in a newer one. |
+| Sessions in `.luxe/sessions/` | A newer luxe reads every older session. An older luxe skips what it does not know in a newer one. |
| The RPC protocol | Adding to it changes nothing. Removing or renaming anything changes `protocol_version`. |
| Hook points | Added, never renamed. An unknown point skips that hook alone, with a warning in the log. |
| Tool arguments | A renamed argument still accepts its old name. |
@@ -1498,9 +1501,9 @@ Before 1.0, a minor release may change the RPC protocol and the hook points, and
will say so. Config files and sessions keep the stronger promise regardless, since they are the two
you cannot regenerate.
-## What Luxe will not add
+## What luxe will not add
-- **Plugins loaded into Luxe.** A plugin would have all of Luxe's access, however it was sandboxed.
+- **Plugins loaded into luxe.** A plugin would have all of luxe's access, however it was sandboxed.
- **A network listener for RPC.** It would turn a local program into a service that needs
authentication.
@@@ page: github
@@ -1508,11 +1511,11 @@ you cannot regenerate.
title: "Git and GitHub"
summary: "Worktrees, issues, opening and merging pull requests, review threads, and stacks."
order: 7
-group: "Using Luxe"
+group: "Using luxe"
---
Everything that talks to GitHub goes through the GitHub CLI, `gh`, signed in with your own account
-(`gh auth login`). Without it, the rest of Luxe works and these features say why they cannot.
+(`gh auth login`). Without it, the rest of luxe works and these features say why they cannot.
## Work in a worktree
@@ -1541,7 +1544,7 @@ worktree stays out of `git status`.
luxe --issue 123
```
-Luxe reads issue #123, makes a worktree on a branch named for it (`123-short-title`), and starts a
+luxe reads issue #123, makes a worktree on a branch named for it (`123-short-title`), and starts a
session there. The issue and its latest comments go to the agent with your first message, so "fix
it" is enough. They are treated as the reporter's words to weigh, not as instructions.
@@ -1558,13 +1561,13 @@ luxe --from-pr 142 # resume the session already working on it
A plain `git push` from the worktree updates the pull request, including one from a fork. A fork's
branch that has to be renamed here, such as its `main` checked out as `mara/main`, gets a remote of
its own (`mara-push-main`) for that push, so pushes from the fork's other branches are not
-redirected. Luxe refuses a pull request that is merged or closed, that your account cannot push to,
+redirected. luxe refuses a pull request that is merged or closed, that your account cannot push to,
or whose author turned off edits by maintainers.
## Open a pull request
Review the work with [Guided Review](guided-review.md), finish (`alt+enter`), and choose **Open pull
-request**. Luxe drafts the title and description from the review, with `Closes #123` when the
+request**. luxe drafts the title and description from the review, with `Closes #123` when the
session started from an issue.

@@ -1622,7 +1625,7 @@ you can see a match before you search for it.

-A changed tab is marked with `*` and keeps your search until you close Luxe. `x` puts back the
+A changed tab is marked with `*` and keeps your search until you close luxe. `x` puts back the
tab's own search, and so does running an empty search.
The tabs always stay in this repository. A `repo:`, `org:` or `user:` you type is ignored.
@@ -1714,9 +1717,9 @@ After the bottom pull request is squash-merged, keep its branch until `/stack` h
one. Deleting it through the API, with `gh pr merge --delete-branch`, makes GitHub close the pull
requests based on it.
-## How often Luxe checks GitHub
+## How often luxe checks GitHub
-Luxe reads the pull request after anything that could change it, such as a push or `/pr`, and on
+luxe reads the pull request after anything that could change it, such as a push or `/pr`, and on
an interval otherwise: every 15 seconds while checks run, every minute once things settle, and every
five minutes when the terminal has been in the background for a while. Each read costs one point
of the 5,000 an hour your `gh` sign-in allows.
@@ -1732,7 +1735,7 @@ your own config only, not a project's.
title: "Guided Review"
summary: "Read a change chapter by chapter, comment for the agent, and finish with a commit, fixes or a pull request."
order: 6
-group: "Using Luxe"
+group: "Using luxe"
---
Guided Review reads a change back to you in the order it makes sense, split into chapters. Each
@@ -1743,9 +1746,9 @@ the agent as work to do.
## Open a review
-Press `ctrl+g`, or run `/diff`. Luxe writes the guide first; when `review ready` shows above the
+Press `ctrl+g`, or run `/diff`. luxe writes the guide first; when `review ready` shows above the
composer, press `ctrl+g` again to open it. Asked for while a turn is running, the review waits for
-the turn to end. A review you already have opens at once, and one you had open when Luxe stopped
+the turn to end. A review you already have opens at once, and one you had open when luxe stopped
comes back with your comments.
| Review | Command |
@@ -1758,7 +1761,7 @@ comes back with your comments.
Inside a review, `s` changes what it covers: uncommitted changes, the last turn, a base or a commit.
A pull request takes `/diff pr`. If the agent committed its work, so nothing is uncommitted,
-`ctrl+g` and `/diff` review the last turn instead. Luxe remembers what you have already reviewed in
+`ctrl+g` and `/diff` review the last turn instead. luxe remembers what you have already reviewed in
each project, so the next review leads with what is new since then.
## Read it
@@ -1805,7 +1808,7 @@ To leave project bookkeeping out of every review:
```
The patterns take `*`, `**`, `?` and a trailing `/`, and a project's list replaces your own.
-Excluded files are counted in the guide. Luxe's own `.luxe/` is never reviewed.
+Excluded files are counted in the guide. luxe's own `.luxe/` is never reviewed.
A review leaves out a file it cannot hold, so the rest is still reviewable: in the last turn, any
file over 1 MiB, unreadable, or not a regular file; in uncommitted changes or a base, a new file
@@ -1832,7 +1835,7 @@ so you can verify it, at any time.
The agent ends a fix by reporting on each comment: fixed, not fixed, or needs clarification, with a
note. The note shows under the comment, and a comment it calls fixed waits for you to verify it,
-since a claim is not proof. If the agent leaves its report out, Luxe asks for it once. A comment
+since a claim is not proof. If the agent leaves its report out, luxe asks for it once. A comment
the agent still says nothing about reads `no word from the agent`, with the reason in Feedback,
and stays where you left it.
@@ -1886,8 +1889,8 @@ agent opened. It needs the GitHub CLI, `gh`, signed in.
| `enter` | Post it |
Line comments are listed under their files with the diff they sit on, the way GitHub shows them.
-Questions you asked the agent stay in Luxe and are never posted. On your own pull request, GitHub
-only takes a comment, so that is all Luxe offers.
+Questions you asked the agent stay in luxe and are never posted. On your own pull request, GitHub
+only takes a comment, so that is all luxe offers.
A drafted summary opens in the preview for you to read and edit. Anything the agent says beside it,
such as a part of the change it did not read, appears under it as a note. Once the review is
@@ -1896,7 +1899,7 @@ posted, the guide's foot says **Posted to GitHub** with the verdict, and `o` ope
### Fix it yourself
To change someone else's pull request rather than comment on it, press `alt+u` (or `w` in the post
-preview). Luxe asks GitHub whether you can push to it, then checks it out in a worktree of its own
+preview). luxe asks GitHub whether you can push to it, then checks it out in a worktree of its own
and starts a session there, with your comments as the agent's first task. A plain `git push` then
updates the pull request.
@@@ page: hooks
@@ -1937,7 +1940,7 @@ case "$cmd" in
esac
```
-A project's `.luxe/hooks.json` runs only when you start Luxe with `--trust-project`, since opening a
+A project's `.luxe/hooks.json` runs only when you start luxe with `--trust-project`, since opening a
repository should never run its commands. When both are loaded, a project's hooks run after yours.
### Hook settings
@@ -1951,7 +1954,7 @@ repository should never run its commands. When both are loaded, a project's hook
`on_timeout: "deny"` makes a guard fail closed. `"ask"` puts the question to you instead, at
`BeforeToolUse`; at other points it acts as `allow`. A
-`decision` Luxe does not recognize, a typo or different casing included, gets the same treatment,
+`decision` luxe does not recognize, a typo or different casing included, gets the same treatment,
with a warning in the log, and its `args` and `context` are dropped.
## Matchers
@@ -1976,10 +1979,10 @@ points without a tool call ignore the matcher.
| Point | Runs | stdin | Can |
| --- | --- | --- | --- |
| `SessionStart` | A session starts or resumes, including a switch to it from another session | `{"Other": null}` | |
-| `SessionEnd` | A session ends, including a switch to another session; Luxe waits for the hook | `{"Other": null}` | |
+| `SessionEnd` | A session ends, including a switch to another session; luxe waits for the hook | `{"Other": null}` | |
| `UserPromptSubmit` | You send a message, including one sent while the agent works | `{"Other": {"content": "…"}}` | block it, add context |
| `BeforeAgentStart` | A turn is about to start | `{"Other": null}` | |
-| `AgentStart` | Luxe is about to call the model | `{"Other": null}` | |
+| `AgentStart` | luxe is about to call the model | `{"Other": null}` | |
| `BeforeToolUse` | Before each tool call | `{"ToolUse": {"tool_name": "bash", "args": {…}}}` | block it, ask you, rewrite it |
| `AfterToolUse` | After each tool call | `{"ToolResult": {"tool_name": "bash", "result": "…", "is_error": false}}` | add context to the result |
| `FileChanged` | `write`, `edit`, an applied rename or code action, or an `ast_grep` rewrite changed a file | `{"FileChanged": {"path": "/abs/file", "operation": "edit"}}` | |
@@ -2001,7 +2004,7 @@ that hook alone, with a warning naming the nearest real point.
| Field | Where | Does |
| --- | --- | --- |
-| `decision` | Required in any answer | `deny` blocks the call or message. `ask` raises a permission prompt, at `BeforeToolUse`. `allow`, or no output at all, means no opinion. An answer without `decision` counts as one Luxe cannot read. |
+| `decision` | Required in any answer | `deny` blocks the call or message. `ask` raises a permission prompt, at `BeforeToolUse`. `allow`, or no output at all, means no opinion. An answer without `decision` counts as one luxe cannot read. |
| `reason` | with `deny` or `ask` | Shown to the model when a call is blocked, to you when a message is blocked, and on the prompt |
| `args` | `BeforeToolUse` | New values for the call's arguments. Keys you leave out keep their values. |
| `context` | `AfterToolUse`, `UserPromptSubmit` | Text added to the tool result, or saved with your message, for the model to see |
@@ -2054,7 +2057,7 @@ saved. Read a hook script before installing it, as you would anything on your `P
}
```
-Luxe already adds language server errors after every edit; a hook like this is for a different
+luxe already adds language server errors after every edit; a hook like this is for a different
check.
**Refuse messages that contain an internal token:**
@@ -2078,7 +2081,7 @@ command is destructive.
## Coming from Claude Code
-| Claude Code | Luxe |
+| Claude Code | luxe |
| --- | --- |
| `PreToolUse`, `PostToolUse` | `BeforeToolUse`, `AfterToolUse` |
| `UserPromptSubmit` with `additionalContext` | `UserPromptSubmit` with `context` |
@@ -2092,16 +2095,16 @@ command is destructive.
@@@ page: index
---
title: "Overview"
-summary: "What Luxe is, and where to find what you need."
+summary: "What luxe is, and where to find what you need."
order: 0
group: "Get started"
---
-Luxe is a coding agent for your terminal. It works in your project with the model you choose,
+luxe is a coding agent for your terminal. It works in your project with the model you choose,
then walks you through what it changed so you can review it and open a pull request without leaving
the terminal.
-New to Luxe? [Install](install.md) it, then follow the [Quickstart](quickstart.md). The rest of the
+New to luxe? [Install](install.md) it, then follow the [Quickstart](quickstart.md). The rest of the
guide is reference, for when you need it.
## Find what you need
@@ -2117,10 +2120,10 @@ guide is reference, for when you need it.
| Resume a session, or see what fills the context | [Sessions and context](sessions.md) |
| See what the agent can use | [Tools](tools.md) |
| Let the agent work through a plan on its own | [Autonomous runs](loop.md) |
-| Reach your sessions from your phone or tablet | [Luxe on your phone](remote.md) |
+| Reach your sessions from your phone or tablet | [luxe on your phone](remote.md) |
| Change the theme, keys or statusline | [Configuration](configuration.md) |
| Give the agent your project's conventions | [Project context](project-context.md) |
-| Add tools, skills or your own commands | [Extending Luxe](extending.md), [Recipes](recipes.md) |
+| Add tools, skills or your own commands | [Extending luxe](extending.md), [Recipes](recipes.md) |
| Control what the agent may do | [Security and permissions](security.md) |
| Look up a command, flag or environment variable | [Command line](cli.md), [Environment variables](environment.md) |
| Fix something that is not working | [Troubleshooting](troubleshooting.md) |
@@ -2128,14 +2131,14 @@ guide is reference, for when you need it.
## Help
- **Help** (`/help`) lists every key for the screen you are on, with any you have rebound.
-- `luxe doctor` checks which language servers, debuggers and tools Luxe can use on your machine.
+- `luxe doctor` checks which language servers, debuggers and tools luxe can use on your machine.
- [Troubleshooting](troubleshooting.md) covers common problems and how to report a bug.
- Problems and questions: [GitHub issues](https://github.com/evanwhittaker/luxe/issues).
-## Ask Luxe
+## Ask luxe
-This guide is built into Luxe, for the version you have installed. Ask the agent how to do
-something in Luxe, such as *how do I add an MCP server?*, and it reads the answer from the guide
+This guide is built into luxe, for the version you have installed. Ask the agent how to do
+something in luxe, such as *how do I add an MCP server?*, and it reads the answer from the guide
rather than guessing. You can read it yourself too:
```bash
@@ -2149,11 +2152,11 @@ luxe docs --search "on_ask"
Every page is also plain markdown, at the same address with `.md` on the end:
`https://luxe.dev/docs/install.md`. [`llms.txt`](https://luxe.dev/llms.txt) lists every page with a
one-line summary, and [`llms-full.txt`](https://luxe.dev/llms-full.txt) is the whole guide in one
-file. Inside Luxe, `luxe docs` serves the same pages offline.
+file. Inside luxe, `luxe docs` serves the same pages offline.
@@@ page: install
---
title: "Install"
-summary: "Install Luxe, check your setup, and pick a font."
+summary: "Install luxe, check your setup, and pick a font."
order: 1
group: "Get started"
---
@@ -2177,6 +2180,11 @@ With Homebrew, on macOS or Linux:
brew install evanwhittaker/tap/luxe
```
+Homebrew finishes with a line that names the installed version; run `luxe --version`
+to check. Since Homebrew 6, a tap has to be trusted before its code runs. The full name above
+trusts luxe's formula for you, but Homebrew can still stop on *another* tap you already have. If it
+names one, trust it with `brew trust /`, or remove it with `brew untap`.
+
With Cargo, which builds it, so you need Rust 1.89 or newer ([rustup](https://rustup.rs) installs
one):
@@ -2196,7 +2204,7 @@ about your policy is saved.
Any of them installs the `luxe` command; `luxe --version` checks it worked.
-To update, run `luxe update`, or `/update` inside Luxe. Luxe checks once a day for a newer release
+To update, run `luxe update`, or `/update` inside luxe. luxe checks once a day for a newer release
and says so on the start screen; see [Updates](configuration.md#updates) for how it updates and how
to turn the check off.
@@ -2220,17 +2228,17 @@ cargo build --release
## Requirements
-Luxe runs natively on Linux, macOS and Windows. Only a terminal and a model are required.
+luxe runs natively on Linux, macOS and Windows. Only a terminal and a model are required.
| | |
| --- | --- |
| **Terminal** | Any current terminal with UTF-8. On Windows, use Windows Terminal. |
| **Model** | An API key, a ChatGPT, Copilot or OpenRouter subscription, or a local server such as LM Studio or Ollama. See [Providers](providers.md). |
-| **Git** | For reviewing changes and working in worktrees. Without a repository, Luxe runs with nothing to review. |
+| **Git** | For reviewing changes and working in worktrees. Without a repository, luxe runs with nothing to review. |
| **GitHub CLI** | Optional. `gh`, signed in, for issues and pull requests. |
| **A Nerd Font** | Optional, for icons. See [Fonts](#fonts). |
-Luxe also uses these when they are installed:
+luxe also uses these when they are installed:
- `ast-grep`, for structural search.
- A language server per language, for code intelligence. See [Code
@@ -2243,7 +2251,7 @@ Luxe also uses these when they are installed:
luxe doctor
```
-`luxe doctor` lists the tools, language servers and debug adapters Luxe found, which ones work, and
+`luxe doctor` lists the tools, language servers and debug adapters luxe found, which ones work, and
how to install what is missing:
```
@@ -2259,7 +2267,7 @@ Language servers
## Fonts
-Luxe draws its icons from a [Nerd Font](https://www.nerdfonts.com/) when your terminal has one.
+luxe draws its icons from a [Nerd Font](https://www.nerdfonts.com/) when your terminal has one.
Without one, it falls back to plain characters on its own, and every icon has a word beside it, so
nothing becomes unreadable.
@@ -2287,7 +2295,7 @@ desktop's font instead. To check what a name resolves to:
fc-match -f '%{file}\n' "JetBrainsMono NFM"
```
-JetBrains Mono lacks six symbols Luxe draws (`ℹ︎ ↳︎ ↺︎ ⏸︎ ✔︎ ✖︎`), and your system fills them in
+JetBrains Mono lacks six symbols luxe draws (`ℹ︎ ↳︎ ↺︎ ⏸︎ ✔︎ ✖︎`), and your system fills them in
from another font. For every character from one face, use **IosevkaTerm** or **Adwaita Mono**:
```bash
@@ -2298,26 +2306,26 @@ scoop install nerd-fonts/IosevkaTerm-NF-Mono nerd-fonts/AdwaitaMono-NF-Mono
```
To use the wider icons anyway, set `"glyphs": "nerd_wide"` in
-[`config.json`](configuration.md), and Luxe gives each icon two cells.
+[`config.json`](configuration.md), and luxe gives each icon two cells.
## On Windows
-Install [Git for Windows](https://gitforwindows.org/). Luxe finds its `bash` even though it is not
+Install [Git for Windows](https://gitforwindows.org/). luxe finds its `bash` even though it is not
on `PATH`, and the agent needs a POSIX shell for most commands. Without one, commands run through
`cmd.exe`, which handles `cargo test` or `npm run build` but not much else.
-If WSL is installed, Luxe skips the `bash.exe` in `System32`, which would run commands inside Linux
+If WSL is installed, luxe skips the `bash.exe` in `System32`, which would run commands inside Linux
against a different filesystem.
Python on Windows is `py` or `python`. `python3` is usually the Microsoft Store's placeholder, so
-Luxe tries `py -3` first. For the Python debugger, run `py -m pip install debugpy`. The agent's
+luxe tries `py -3` first. For the Python debugger, run `py -m pip install debugpy`. The agent's
commands run Python in UTF-8 mode (`PYTHONUTF8=1`) unless you have set it yourself, so output with
characters outside your code page doesn't crash them.
-Windows Terminal keeps a column at the right edge for its scrollbar. To give it to Luxe, set
+Windows Terminal keeps a column at the right edge for its scrollbar. To give it to luxe, set
**Scrollbar visibility** to **Hidden** in your profile's **Appearance** settings.
-## Where Luxe keeps files
+## Where luxe keeps files
| | Linux and macOS | Windows |
| --- | --- | --- |
@@ -2327,10 +2335,10 @@ Windows Terminal keeps a column at the right edge for its scrollbar. To give it
An `XDG_CONFIG_HOME`, `XDG_STATE_HOME` or `XDG_CACHE_HOME` you have set wins on every platform.
`LUXE_CONFIG_DIR` moves every configuration file, API key and provider sign-in. `LUXE_STATE_DIR`
-moves the state Luxe keeps for you: the model you last chose, MCP sign-ins, and the logs of
+moves the state luxe keeps for you: the model you last chose, MCP sign-ins, and the logs of
commands that only report, such as `luxe auth`.
-Inside a project, Luxe keeps its sessions, logs and caches in `.luxe/`, with a `.gitignore` that
+Inside a project, luxe keeps its sessions, logs and caches in `.luxe/`, with a `.gitignore` that
keeps all of it out of Git except the files a team may want to share: `permissions.json`,
`mcp.json`, `hooks.json` and `skills/`.
@@@ page: interface
@@ -2338,7 +2346,7 @@ keeps all of it out of Git except the files a team may want to share: `permissio
title: "The interface"
summary: "The start screen, the conversation, the composer, the dock, the statusline and the preview."
order: 3
-group: "Using Luxe"
+group: "Using luxe"
---

@@ -2350,7 +2358,7 @@ This page names the keys for each part of the screen. All of them, and how to re
## The start screen
-Luxe opens on a prompt. Type and press `enter` to ask in a new session, or choose a row from the
+luxe opens on a prompt. Type and press `enter` to ask in a new session, or choose a row from the
list under it with `↑`/`↓` and `enter`.
| Row | Opens |
@@ -2361,7 +2369,7 @@ list under it with `↑`/`↓` and `enter`.
| **Sessions…** | Every session in this project, searchable. `/sessions` opens it too, here or in a session. |
| **Review changes** | A [Guided Review](guided-review.md) of your uncommitted changes, when there are any |
| **Configuration** | Providers, credentials, [settings](configuration.md#settings) and config files, once a provider is set up |
-| **Quit** | Leaves Luxe. `ctrl+c` does too, here and in a session, even mid-turn, unless a prompt or menu is open, which it answers instead. |
+| **Quit** | Leaves luxe. `ctrl+c` does too, here and in a session, even mid-turn, unless a prompt or menu is open, which it answers instead. |
| **RECENT** | Your other recent sessions |
Typing narrows the list. The first row becomes **Ask in a new session**, and under it come the
@@ -2395,7 +2403,7 @@ screen and opens a new session. **Start screen** in `/settings` sets the same ke

-Luxe works on the project it was started in. To switch projects, `cd` there and run `luxe`.
+luxe works on the project it was started in. To switch projects, `cd` there and run `luxe`.
## The conversation
@@ -2494,7 +2502,7 @@ conversation, or in its place below 80 columns.
`mode` is `auto` (beside the conversation when there is room), `shown` or `hidden`.
Drag the line between the dock and the conversation to make the dock wider or narrower. It stays
-that width the next time Luxe starts; double-click the line to put it back.
+that width the next time luxe starts; double-click the line to put it back.
### Files
@@ -2600,12 +2608,12 @@ scrollbar to scroll, or press anywhere on it to jump.
The lines between panes move: the dock's, the preview's, and the one beside a review's guide. A line
lights under the pointer; drag it to resize the pane beside it, or double-click it to put the pane
-back to its own width. Luxe remembers the widths for next time.
+back to its own width. luxe remembers the widths for next time.
Drag across the conversation or the composer to select, and the selection is copied when you let
go. Click a message to copy all of it. `ctrl+insert` or `ctrl+shift+c` copies a selection too.
-Your terminal's own copy key reaches Luxe only if the terminal passes it through. Either hold
+Your terminal's own copy key reaches luxe only if the terminal passes it through. Either hold
`shift` (`option` on macOS) while dragging, to make a selection the terminal copies itself, or let
the copy key fall through when the terminal has nothing selected. In Ghostty:
@@ -2613,7 +2621,7 @@ the copy key fall through when the terminal has nothing selected. In Ghostty:
keybind = performable:control+insert=copy_to_clipboard
```
-Luxe copies to the desktop clipboard, and over SSH through the terminal (OSC 52). Inside tmux,
+luxe copies to the desktop clipboard, and over SSH through the terminal (OSC 52). Inside tmux,
that needs `set -g set-clipboard on`.
`/mouse off` hands the mouse back to the terminal for the session, and `"mouse": false` in
@@ -2624,7 +2632,7 @@ notch scrolls, from 1 to 20 lines (3 by default).
title: "Autonomous runs"
summary: "Let the agent work through a phased plan on its own, with tests and a reviewer at each phase."
order: 12
-group: "Using Luxe"
+group: "Using luxe"
---
`luxe loop` runs the agent on its own, phase by phase through your plan. A phase only counts as done
@@ -2677,7 +2685,7 @@ A run uses your session's model, or `--model`. Only one run at a time works in a
## Check each phase
-When the agent says a phase is done, Luxe checks it before moving on:
+When the agent says a phase is done, luxe checks it before moving on:
1. **Your commands.** Each must pass.
2. **A reviewer**, if `gate.review` is on and the commands passed. It runs in a fresh session, so
@@ -2720,13 +2728,22 @@ phase with `luxe loop phase --set `.
A run prints one line per step, timed from its start:
```text
- 19s write logstats.py
- 49s test python3 -m unittest -q
- 1m09s phase 0 claimed complete — running the gate
- 1m31s the acceptance commands passed — asking the reviewer
- 2m34s phase 0 accepted — now on phase 1
+ 0s ┃ phase 0
+ 0s ┃ ┌ start
+ 19s ┃ │ ✓ write logstats.py
+ 49s ┃ │ ✗ test python3 -m unittest -q · failed
+ 1m09s ┃ │ phase 0 claimed complete — running the gate
+ 1m31s ┃ │ the acceptance commands passed — asking the reviewer
+ 2m34s ┃ │ phase 0 accepted — now on phase 1
+ 2m40s ┃ └ turn ended · 6 tools · 2m40s
+ 2m41s ┃ phase 1
+ 2m41s ┃ ┌ iteration 1
```
+The heavy rail on the left marks the phase, in a color of its own, and the bracket inside it marks
+each iteration. In a terminal, a tool's line shows `○` while it runs, then turns green with `✓` or
+red with `✗` when it ends. Piped to a file there's no color, and a failure gets a line of its own.
+
`luxe loop status` shows what it has cost, by model:
```text
@@ -2760,7 +2777,7 @@ group: "Extend"
---
An [MCP](https://modelcontextprotocol.io) server gives the agent new tools: a browser, a database,
-your issue tracker. Luxe reaches every server through one `mcp` tool, so a server's tools cost
+your issue tracker. luxe reaches every server through one `mcp` tool, so a server's tools cost
nothing in a request until the agent looks for them:
```text
@@ -2798,14 +2815,14 @@ removed by editing that file. `luxe mcp list` starts nothing: it shows every con
disabled ones included, and marks a project's entries `not started (project)`, with or without
`--trust-project`. In a session, `/mcp` shows each server, how it connects and its tools.
-Luxe also reads `~/.config/mcp/mcp.json`, and a project's `.mcp.json` and `.luxe/mcp.json`. Where
+luxe also reads `~/.config/mcp/mcp.json`, and a project's `.mcp.json` and `.luxe/mcp.json`. Where
two files name the same server, the project's wins.
### Server settings
| Key | Does |
| --- | --- |
-| `command`, `args`, `env` | Start a local server. An `env` value can use `${VAR}` or `$env:VAR`; an unset one is empty, and Luxe logs which. |
+| `command`, `args`, `env` | Start a local server. An `env` value can use `${VAR}` or `$env:VAR`; an unset one is empty, and luxe logs which. |
| `url`, `headers` | Connect to a remote server. `headers` take `${VAR}` the same way. |
| `auth` | Sign in: see [below](#sign-in) |
| `lifecycle` | `lazy` (the default) connects on first use and disconnects when idle. `eager` connects at start; `keep_alive` stays connected. |
@@ -2829,7 +2846,7 @@ A remote server signs in with a token or with OAuth.
}
```
-For OAuth, sign in once in a browser, and Luxe refreshes the token from then on:
+For OAuth, sign in once in a browser, and luxe refreshes the token from then on:
```bash
luxe mcp login --id linear
@@ -2837,7 +2854,7 @@ luxe mcp login --id linear
The agent never starts a sign-in. Tokens are kept in `~/.local/state/luxe/mcp-auth.json`, readable
only by you. With `LUXE_STATE_DIR` set they are kept there instead, so a moved state directory does
-not share the default one's sign-ins. If that file stops being valid JSON, Luxe moves it to
+not share the default one's sign-ins. If that file stops being valid JSON, luxe moves it to
`mcp-auth.json.bad` (or a dated name, if one is already there) rather than write over it, and you
sign in again.
@@ -2910,7 +2927,7 @@ A server's descriptions and results are treated as information, never as instruc
### Servers a project brings
-A project's `.mcp.json` names programs that would run on your machine. Luxe starts a project's
+A project's `.mcp.json` names programs that would run on your machine. luxe starts a project's
servers in an interactive session, or with `--trust-project`, but not in a headless or loop run. To
decide for every run, set `lsp.project_servers` in your own config; see
[Tools](tools.md#a-server-the-project-installed).
@@ -2950,14 +2967,14 @@ row, and the agent receives the text your program gave for it.
| `icon` | One Nerd Font glyph for the tab, as the character or a JSON escape. Without one, the tab shows a tag; without a Nerd Font, a plain mark. |
Sources are read from your own config only, since each one runs a program. Their tabs come after
-Luxe's own: Files, Skills, GitHub and MCP, whose kinds `file`, `skill`, `github` and `mcp` are
+luxe's own: Files, Skills, GitHub and MCP, whose kinds `file`, `skill`, `github` and `mcp` are
taken. Two sources with the same kind get one tab, the first's.
## What the program prints
-Luxe runs the program, in the project, the first time its tab opens, and lists what it prints: one
-JSON object per line. The rows are kept until Luxe exits, so a new ticket, or a change to your
-script, shows up the next time you start Luxe. The program has no terminal and no stdin, so it
+luxe runs the program, in the project, the first time its tab opens, and lists what it prints: one
+JSON object per line. The rows are kept until luxe exits, so a new ticket, or a change to your
+script, shows up the next time you start luxe. The program has no terminal and no stdin, so it
cannot prompt for a sign-in.
```json
@@ -2977,7 +2994,7 @@ wrong. If the program fails, the tab shows the last line it wrote to stderr, or
and it runs again the next time the tab opens.
Make `text` enough for the agent to find the rest: a title and a link, or an ID it can look up with
-a CLI or an MCP server. A mention you type by hand, or one not picked since Luxe started, reaches
+a CLI or an MCP server. A mention you type by hand, or one not picked since luxe started, reaches
the agent as `Jira PROJ-12` alone. What a tracker says is other people's words, and it reaches your model
provider once you mention it.
@@ -2998,13 +3015,13 @@ jq -c --arg site "$site" '.issues[] | {
```
Make it executable (`chmod +x`), put it on your `PATH`, and add the Jira entry above. For Asana,
-see [Recipes](recipes.md#asana). Or ask Luxe to write one for the tracker you use.
+see [Recipes](recipes.md#asana). Or ask luxe to write one for the tracker you use.
@@@ page: mentions
---
title: "Mentions"
summary: "Point the agent at a file, skill, issue, MCP server or your own tracker with `@`."
order: 5
-group: "Using Luxe"
+group: "Using luxe"
---
Type `@` at the start of a word in the composer to mention something. The picker opens on your
@@ -3054,7 +3071,7 @@ In `/github`, `c` inserts a mention of the pull request or issue under the curso
@@@ page: project-context
---
title: "Project context"
-summary: "Give the agent your project's conventions in AGENTS.md, and how Luxe finds the project."
+summary: "Give the agent your project's conventions in AGENTS.md, and how luxe finds the project."
order: 3
group: "Customize"
---
@@ -3065,12 +3082,12 @@ Put your project's conventions in an `AGENTS.md` at its root, and the agent read
every session. In a project folder a `CLAUDE.md` works too; when both exist, `AGENTS.md` is used.
The global files below are read only by those names, so `~/.claude/CLAUDE.md` is not.
-Luxe reads, in order:
+luxe reads, in order:
1. `~/.agents/AGENTS.md`, the file many people keep for every agent they use
-2. `~/.config/luxe/agents.md`, for every project in Luxe
+2. `~/.config/luxe/agents.md`, for every project in luxe
3. `AGENTS.md` at the top of the repository
-4. One `AGENTS.md` per folder down to where you started Luxe
+4. One `AGENTS.md` per folder down to where you started luxe
So a monorepo can keep shared rules at the root and package rules beside the code, and `luxe`
started in `packages/app` reads both. Nothing above the repository is read. Outside a repository,
@@ -3086,7 +3103,7 @@ agent is told with your next message.
An `AGENTS.md` in a folder below the repository's top, but not on the way to where you started,
joins when the agent first opens, writes or searches a path in that folder. It's added to that
-tool result, once, and Luxe shows **Following** and the file's path. The system prompt doesn't
+tool result, once, and luxe shows **Following** and the file's path. The system prompt doesn't
change, so the cache holds. Ignored folders, such as `node_modules`, are skipped. If a rewind or a
compaction drops the file, it joins again the next time the agent goes there.
`LUXE_NESTED_INSTRUCTIONS=0` turns this off.
@@ -3094,7 +3111,7 @@ compaction drops the file, it joins again the next time the agent goes there.
**Project context** lists every file a turn reads, and says whose rules each one carries:
**global**, **project** or **folder** context. Type `context` on the start screen and open it.
`enter` opens a file in your editor, or creates one from a template. The template's guidance is in
-comments, which Luxe removes and reports, so replace them with your own lines.
+comments, which luxe removes and reports, so replace them with your own lines.
### Duplicates
@@ -3130,7 +3147,7 @@ lists and architecture overviews do not; the agent reads those from the code.
An `AGENTS.md` over 24,000 characters is cut off, with a note telling the agent to read the rest of
the file itself.
-### What Luxe removes
+### What luxe removes
Two kinds of text are removed from `AGENTS.md` and skill files before the agent sees them, because
neither is visible to someone reviewing the file:
@@ -3141,13 +3158,13 @@ neither is visible to someone reviewing the file:
and soft hyphens and other marks that draw nothing, written as characters or as references
such as ``
-A file that repeats one link label so often that Luxe cannot check every definition is left out
-entirely. For an `AGENTS.md`, Luxe says what it removed or left out when it starts; a skill file is
+A file that repeats one link label so often that luxe cannot check every definition is left out
+entirely. For an `AGENTS.md`, luxe says what it removed or left out when it starts; a skill file is
cleaned without a notice.
## Where the project begins
-Luxe works on one project, found by walking up from where you started it:
+luxe works on one project, found by walking up from where you started it:
1. Inside a repository (`.git`, `.jj` or `.hg`), the nearest folder with a `.luxe/` in it, up to
the top of the repository
@@ -3165,14 +3182,14 @@ mkdir -p packages/app/.luxe
The package keeps its own sessions and still reads the `AGENTS.md` at the top of the repository.
-Outside a repository, a `.luxe/` claims only its own folder. Luxe writes one wherever it runs, so
+Outside a repository, a `.luxe/` claims only its own folder. luxe writes one wherever it runs, so
starting it once in `~/.config` does not make `~/.config/herdr` part of that project.
@@@ page: providers
---
title: "Models and providers"
summary: "Sign in to a provider, choose a model and thinking level, and run local models."
order: 9
-group: "Using Luxe"
+group: "Using luxe"
---
## Sign in
@@ -3239,7 +3256,7 @@ used from other programs.
GitHub Copilot serves Claude through Anthropic's Messages API, its newer GPT, Grok and MAI models
only through its Responses API, and the rest through chat completions. Copilot says which model
-takes which, and Luxe sends each one where it belongs. Claude goes to Messages even though chat
+takes which, and luxe sends each one where it belongs. Claude goes to Messages even though chat
completions would take it, because only Messages keeps its thinking, its effort level and the
prompt cache for the whole conversation. The models offered are your plan's own, kept current each
time the sign-in refreshes. Signing in turns on any model your plan offers that you have not been
@@ -3263,8 +3280,9 @@ A command runs each time a request needs the key.
## Choose a model
`/model` opens the model picker: every model your providers can serve, configured providers first.
-Type to search. The session carries on with the new model, even across providers, and Luxe
-remembers your choice for next time.
+Type to search. The session carries on with the new model, even across providers, and luxe
+remembers your choice for next time, in this project and as your latest anywhere
+([which model a session starts on](agent.md#which-model-a-session-starts-on)).
From the command line, `--model` takes a model ID, or `provider/id` to pick the provider too:
@@ -3272,7 +3290,7 @@ From the command line, `--model` takes a model ID, or `provider/id` to pick the
luxe --model anthropic/claude-opus-5
```
-To see what Luxe knows about a model, or list what you can use:
+To see what luxe knows about a model, or list what you can use:
```bash
luxe models show anthropic/claude-opus-5 # context window, output limit, thinking levels, price
@@ -3280,8 +3298,8 @@ luxe models list --provider anthropic # every model; * is active, + is conf
luxe models refresh --force # update the model catalog now
```
-Luxe ships with the [models.dev](https://models.dev) catalog built in and refreshes it in the
-background, so a model released after your copy of Luxe still has the right context window and
+luxe ships with the [models.dev](https://models.dev) catalog built in and refreshes it in the
+background, so a model released after your copy of luxe still has the right context window and
price.
## Thinking
@@ -3290,7 +3308,7 @@ price.
**Medium**, **High**, **XHigh** or **Ultra**. Levels the model does not support are grayed out.
`--thinking high` sets it at launch.
-Luxe remembers the level per model, since `high` on a frontier model and on a small local one cost
+luxe remembers the level per model, since `high` on a frontier model and on a small local one cost
very different amounts of time. To give a model its own starting level:
```json title="~/.config/luxe/models.json"
@@ -3312,7 +3330,7 @@ ChatGPT plan, the statusline shows how much of the plan is used, such as `12% we
shows when it resets. On a Copilot plan it shows the month's allowance the same way, such as `19%
monthly`.
-Luxe prices Copilot usage at Copilot's per-token rates, so the figure matches what Copilot reports
+luxe prices Copilot usage at Copilot's per-token rates, so the figure matches what Copilot reports
for each request. A Copilot plan billed by the token spends its monthly allowance at those rates.
A ChatGPT subscription serves its own set of models, different from OpenAI's API, and `/model`
@@ -3336,7 +3354,7 @@ For reasoning controls, chat templates and sampling settings on a local server,
## Microsoft Foundry
-One Foundry resource and key serve three APIs, which Luxe treats as three providers. Signing in to
+One Foundry resource and key serve three APIs, which luxe treats as three providers. Signing in to
one configures all three:
```bash
@@ -3352,7 +3370,7 @@ choose a model with the provider that serves it:
| `foundry-anthropic` | Claude | `luxe --model foundry-anthropic/claude-sonnet-5` |
| `foundry-completions` | DeepSeek, Grok, Llama, Mistral, Phi and Cohere | `luxe --model foundry-completions/deepseek-v4-pro` |
-`/model` lists what your resource has deployed, not every model Azure offers. Luxe asks the
+`/model` lists what your resource has deployed, not every model Azure offers. luxe asks the
resource for its deployments when you sign in and again each time you open `/model`, and sorts them
onto the right provider by the model each one runs. `luxe models refresh` asks too. If a resource
cannot list its deployments, `/model` shows Azure's catalog marked `unverified`, and you can
@@ -3363,8 +3381,8 @@ Entra ID, see [Custom providers](custom-providers.md#keyless-sign-in-with-micros
## When a provider stalls
-Luxe waits 60 seconds for a provider to start answering, and 180 seconds between parts of a reply.
-If a provider is overloaded, rate limited or unreachable before it has sent anything, Luxe tries
+luxe waits 60 seconds for a provider to start answering, and 180 seconds between parts of a reply.
+If a provider is overloaded, rate limited or unreachable before it has sent anything, luxe tries
again after 2, 4 and 8 seconds, showing `retrying (2/3)` above the composer. Once a reply has
started, a failure ends the turn rather than paying for the reply twice. A bad key or a refused
request is not retried.
@@ -3385,7 +3403,7 @@ group: "Get started"
This walks through one change from start to finish: connect a model, ask for something, review
what the agent did, and open a pull request.
-## 1. Start Luxe in your project
+## 1. Start luxe in your project
```bash
cd your-project
@@ -3397,14 +3415,14 @@ Reviewing changes and opening a pull request need a Git repository. A pull reque
branch other than the one it merges into; `luxe --worktree` makes one for you, in a
[worktree](github.md#work-in-a-worktree) of its own.
-The first time, Luxe asks four questions: a theme, a provider, which tools the agent can use, and
-what Luxe opens on. Each one previews or explains itself, and `enter` moves on. `s` skips setup and
+The first time, luxe asks four questions: a theme, a provider, which tools the agent can use, and
+what luxe opens on. Each one previews or explains itself, and `enter` moves on. `s` skips setup and
keeps none of the answers, though a provider you signed in to stays signed in. Your answers go to
`~/.config/luxe/config.json`, and `/settings` or `luxe init` changes them later.

-After that, Luxe opens on the start screen: a prompt, with your last session, your other sessions
+After that, luxe opens on the start screen: a prompt, with your last session, your other sessions
and **Configuration** listed under it, or **Set up a provider** first if you have none yet. Type to ask something or to find a session or a command, or
use `↑`/`↓` and `enter` to open a row.
@@ -3425,7 +3443,7 @@ and pick one:
When it connects, the model picker opens. Choose a model, and change it any time with `/model`.
-An API key in your environment works too. Luxe finds `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`,
+An API key in your environment works too. luxe finds `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`,
`GEMINI_API_KEY` and the rest, and offers that provider's models. To sign in from a script or a
machine without the interface, use the CLI:
@@ -3458,7 +3476,7 @@ place on a narrow one.
## 4. Review the change
-Press `ctrl+g` to start a **Guided Review**. Luxe writes a guide to the change, split into
+Press `ctrl+g` to start a **Guided Review**. luxe writes a guide to the change, split into
chapters, each with what changed, why, and how it was checked, next to the code. When
`review ready` shows above the prompt, press `ctrl+g` again to open it.
@@ -3477,7 +3495,7 @@ Comments go to the agent as work to do when you finish with **Request fixes**.
## 5. Open a pull request
This needs the GitHub CLI, `gh`, signed in (`gh auth login`): **Open pull request** is offered once
-`gh` says the branch has none. Finish the review (`alt+enter`) and choose it. Luxe drafts the title
+`gh` says the branch has none. Finish the review (`alt+enter`) and choose it. luxe drafts the title
and description from the review. The heading says what happens before it opens, such as
"commits and pushes first". The pull request pictured came from a session that took up an issue
([`luxe --issue 123`](github.md#start-from-an-issue)), so its branch is named for the issue and its
@@ -3614,7 +3632,7 @@ and the agent can look it up through the MCP server. See [Mention sources](menti
## Subagents
A subagent is a separate context that does one job and returns a summary, which is what a headless
-Luxe run is:
+luxe run is:
```bash
luxe --headless --read-only --model openai/gpt-5.6 --format json \
@@ -3637,7 +3655,7 @@ To let the agent start one itself, wrap that command in an MCP server with a
- Return the subagent's stderr when it fails, so the agent sees why.
- Give a subagent that writes its own worktree, with `git worktree add`, so it stays out of yours.
-Luxe refuses to nest more than two levels deep, since an agent that starts two subagents per turn
+luxe refuses to nest more than two levels deep, since an agent that starts two subagents per turn
would multiply fast. `LUXE_MAX_DEPTH` raises the limit.
## A guard that asks a model
@@ -3690,10 +3708,10 @@ fi
- **Nobody answers in a headless run**, so its `ask` is allowed there. Use `--on-ask deny`, or
`"hook_ask": "deny"` in `permissions.json`, to make it hold.
-## Run Luxe in Herdr
+## Run luxe in Herdr
[Herdr](https://herdr.dev) keeps coding agents in panes and shows which ones are working, waiting
-on you, or done. Luxe reports its state to Herdr on its own when it runs in a Herdr pane:
+on you, or done. luxe reports its state to Herdr on its own when it runs in a Herdr pane:
**working** during a turn, **blocked** when it needs you, and **done** when the turn ends.
```bash
@@ -3710,11 +3728,11 @@ curl -fsSL https://raw.githubusercontent.com/ogulcancelik/herdr/master/SKILL.md
-o ~/.config/luxe/skills/herdr/SKILL.md
```
-`LUXE_HERDR=0` or `{ "herdr": false }` in Luxe's config turns the reporting off.
+`LUXE_HERDR=0` or `{ "herdr": false }` in luxe's config turns the reporting off.
-### Luxe colors in Herdr
+### luxe colors in Herdr
-Herdr has no theme files — it takes a built-in theme plus per-token overrides. This is Luxe's
+Herdr has no theme files — it takes a built-in theme plus per-token overrides. This is luxe's
own palette in those tokens, so the pane and the multiplexer around it read as one program:
```toml title="~/.config/herdr/config.toml"
@@ -3722,12 +3740,12 @@ own palette in those tokens, so the pane and the multiplexer around it read as o
name = "vesper" # the closest built-in; every token below is overridden anyway
[theme.custom]
-accent = "#ddc084" # champagne, the one color Luxe leads with
+accent = "#ddc084" # champagne, the one color luxe leads with
text = "#e4dfd6"
subtext0 = "#97938c"
overlay0 = "#4d4a45"
overlay1 = "#97938c"
-sidebar_bg = "#12100e" # the base surface, same as Luxe's background
+sidebar_bg = "#12100e" # the base surface, same as luxe's background
panel_bg = "#12100e" # the tab bar sits on it too, so the chrome is one field
surface0 = "#201e1a"
surface1 = "#2e2b27"
@@ -3739,7 +3757,7 @@ yellow = "#ddc084" # working — an agent at work glows in the accent
green = "#81caa0" # idle / done
red = "#e68485" # blocked, needs you
peach = "#faaa51" # interrupted
-blue = "#85c7d7" # Luxe has one cool hue, so blue and teal share it
+blue = "#85c7d7" # luxe has one cool hue, so blue and teal share it
teal = "#85c7d7"
[ui]
@@ -3747,18 +3765,18 @@ accent = "#ddc084" # borders and navigation chrome, which read off their
```
Reload it in place with `prefix+shift+r`, or `herdr server reload-config`. For any of the other
-[themes](appearance.md) Luxe ships, the hexes are in its theme file and the token names above do
+[themes](appearance.md) luxe ships, the hexes are in its theme file and the token names above do
not change.
The one token worth a second thought is `panel_bg`: Herdr draws the active tab as `panel_bg` text
on an `accent` fill, so it has to stay dark enough to read there. Lifting it to `#201e1a` gives the
tab bar a surface of its own if you prefer the chrome delineated.
-## Luxe from your phone
+## luxe from your phone
Your sessions keep running on your computer, and your phone or tablet reaches them over
[Tailscale](https://tailscale.com), a private network between your own devices. Nothing is opened to
-the internet. This is the terminal over SSH; [Luxe on your phone](remote.md) is Luxe's own way in.
+the internet. This is the terminal over SSH; [luxe on your phone](remote.md) is luxe's own way in.
1. Install Tailscale on the computer and on the phone, and sign in to the same account on both.
2. On the computer, turn on SSH. On Linux and macOS, Tailscale can do it for you:
@@ -3768,10 +3786,10 @@ the internet. This is the terminal over SSH; [Luxe on your phone](remote.md) is
```
On Windows, turn on the OpenSSH Server in Settings, under Optional features.
-3. Run Luxe in [Herdr](#run-luxe-in-herdr), so your sessions outlive the terminal you started
+3. Run luxe in [Herdr](#run-luxe-in-herdr), so your sessions outlive the terminal you started
them in.
4. On the phone, connect with an SSH app such as Blink Shell or Termius, using the computer's
- Tailscale name, and run `herdr`. It attaches to the panes you left, with Luxe where it was.
+ Tailscale name, and run `herdr`. It attaches to the panes you left, with luxe where it was.
```bash
mosh you@desk -- herdr
@@ -3781,24 +3799,24 @@ mosh you@desk -- herdr
cellular. It needs `mosh` installed on the computer too, which Linux and macOS have and Windows
does not; plain `ssh` works everywhere.
-On an iPad with a keyboard, this is Luxe as you know it. On a phone, Luxe fits a narrow screen,
+On an iPad with a keyboard, this is luxe as you know it. On a phone, luxe fits a narrow screen,
and Herdr marks which session is waiting on you.
-### A notification when Luxe needs you
+### A notification when luxe needs you
-[Luxe on your phone](remote.md#notifications) has notifications built in. Without it, a
+[luxe on your phone](remote.md#notifications) has notifications built in. Without it, a
[hook](hooks.md) can post to [ntfy](https://ntfy.sh), whose app shows it on your phone:
```json title="~/.config/luxe/hooks.json"
{
"PermissionRequested": [
- { "command": ["curl", "-s", "-d", "Luxe is waiting for an approval", "https://ntfy.sh/your-topic"] }
+ { "command": ["curl", "-s", "-d", "luxe is waiting for an approval", "https://ntfy.sh/your-topic"] }
],
"BeforeToolUse": [
- { "matcher": "question", "command": ["curl", "-s", "-d", "Luxe has a question", "https://ntfy.sh/your-topic"] }
+ { "matcher": "question", "command": ["curl", "-s", "-d", "luxe has a question", "https://ntfy.sh/your-topic"] }
],
"AgentEnd": [
- { "command": ["curl", "-s", "-d", "Luxe finished", "https://ntfy.sh/your-topic"] }
+ { "command": ["curl", "-s", "-d", "luxe finished", "https://ntfy.sh/your-topic"] }
]
}
```
@@ -3811,11 +3829,11 @@ and Herdr marks which session is waiting on you.
## A local model in LM Studio
1. Load a model in LM Studio and start its server.
-2. In Luxe, open **Configuration** and choose **Connect a local or custom model**, then
+2. In luxe, open **Configuration** and choose **Connect a local or custom model**, then
**LM Studio**. Leave the endpoint blank for the default.
3. Choose the model in the picker that opens.
-LM Studio ignores the chat-template settings Luxe sends a local server by default, so for
+LM Studio ignores the chat-template settings luxe sends a local server by default, so for
thinking levels to reach the model, send them as `reasoning_effort`, and start each session at a
sensible level. Connecting already wrote an `lmstudio` provider to `models.json`, so add these keys
to it rather than replacing the file:
@@ -3834,19 +3852,22 @@ to it rather than replacing the file:
[Custom providers](custom-providers.md) covers samplers and request formats.
@@@ page: remote
---
-title: "Luxe on your phone"
+title: "luxe on your phone"
summary: "Reach the sessions running on your computer from your own phone or tablet, over Tailscale."
order: 13
-group: "Using Luxe"
+group: "Using luxe"
---
`luxe remote` lets your phone or tablet see the sessions running on your computer, join one, and
start new ones. It listens on your computer only. [Tailscale](https://tailscale.com), a private
network between your own devices, carries it to them over HTTPS, so nothing is opened to the
-internet and there is no Luxe server in between.
+internet and there is no luxe server in between.
On the phone it's an app in the browser, drawn like the interface in your theme. For the whole
-terminal, editor included, [Luxe from your phone](recipes.md#luxe-from-your-phone) works over SSH.
+terminal, editor included, [luxe from your phone](recipes.md#luxe-from-your-phone) works over SSH.
+
+
+
## Set it up
@@ -3859,11 +3880,11 @@ terminal, editor included, [Luxe from your phone](recipes.md#luxe-from-your-phon
3. Scan the QR code it shows with the phone's camera, and give the device a name.
4. On an iPhone or iPad, add it to the Home Screen: in Safari, Share, then Add to Home Screen. It
- opens full screen with the Luxe mark as its icon, and keeps the sign-in Safari had when you
+ opens full screen with the luxe mark as its icon, and keeps the sign-in Safari had when you
added it. An app added before pairing asks for the invitation link instead: copy it from
`luxe remote pair` and paste it in.
-Setup takes four steps, and asks before each one that changes something outside Luxe:
+Setup takes four steps, and asks before each one that changes something outside luxe:
| Step | Does |
| --- | --- |
@@ -3886,9 +3907,9 @@ The service is your own, and needs no administrator:
| --- | --- |
| Linux | A systemd user unit, `~/.config/systemd/user/luxe-remote.service` |
| macOS | A launch agent, `~/Library/LaunchAgents/dev.luxe.remote.plist` |
-| Windows | `Luxe remote` under `HKCU\Software\Microsoft\Windows\CurrentVersion\Run` |
+| Windows | `luxe remote` under `HKCU\Software\Microsoft\Windows\CurrentVersion\Run` |
-Without it, run `luxe remote` in a terminal, and your devices reach Luxe while it runs.
+Without it, run `luxe remote` in a terminal, and your devices reach luxe while it runs.
## Devices
@@ -3914,13 +3935,13 @@ luxe remote setup --allow sam@github
Each computer runs its own bridge at its own address, so one phone can reach several: set each up
as above, with Tailscale signed in to the same account, and pair the phone with each. Every one is
its own app on the Home Screen, named for the computer as Tailscale names it (`desk` for
-`desk.tail1234.ts.net`), with the name beside the Luxe mark in its session list and after the
-project in its notifications, "luxe · desk". Luxe finds Tailscale by the `tailscale` command, so
+`desk.tail1234.ts.net`), with the name beside the luxe mark in its session list and after the
+project in its notifications, "luxe · desk". luxe finds Tailscale by the `tailscale` command, so
on a Mac check that `tailscale version` works in Terminal before running setup.
## What your device sees
-Every Luxe running on the computer as you: the interface, `luxe --rpc`, and any started from your
+Every luxe running on the computer as you: the interface, `luxe --rpc`, and any started from your
device. A device that joins a session is a guest, as in [joining a
session](rpc.md#join-a-running-session): it sees the conversation so far and every live event, and
can answer prompts and send messages. It can't switch or end the session, or publish to GitHub.
@@ -3984,8 +4005,8 @@ to `false` is out of reach, from your device too.
## Starting a session
-Tap **New** on the session list to start Luxe in a folder: one where Luxe is running now, one with
-a [herdr](recipes.md#run-luxe-in-herdr) pane, one where the bridge has seen Luxe run, or any path
+Tap **New** on the session list to start luxe in a folder: one where luxe is running now, one with
+a [herdr](recipes.md#run-luxe-in-herdr) pane, one where the bridge has seen luxe run, or any path
you type. What you type narrows the folders it offers, by name and then by path. They're only the
ones above: it never looks through your home folder.
@@ -4036,12 +4057,12 @@ session needs a paired device as well. [Security](security.md#luxe-remote) has t
@@@ page: rpc
---
title: "The RPC protocol"
-summary: "Drive Luxe from your own program: an editor plugin, a dashboard, or a front end of your own."
+summary: "Drive luxe from your own program: an editor plugin, a dashboard, or a front end of your own."
order: 7
group: "Extend"
---
-`luxe --rpc` runs Luxe without its interface and speaks JSON over stdin and stdout, one object per
+`luxe --rpc` runs luxe without its interface and speaks JSON over stdin and stdout, one object per
line. Use it to build an editor plugin, a dashboard, a CI adapter, or a whole new front end, in any
language.
@@ -4051,7 +4072,7 @@ luxe --rpc --github-writes # also lets your client publish to GitHub
```
A client is trusted like the interface it replaces: it answers permission prompts, and it cannot
-loosen any permission rule. Nothing listens on the network. Your client either started Luxe, or
+loosen any permission rule. Nothing listens on the network. Your client either started luxe, or
[joined a session](#join-a-running-session) open in the interface on the same machine.
## A session
@@ -4104,7 +4125,7 @@ for line in luxe.stdout:
elif kind == "turn_ended":
print()
send("close")
- # Ignore everything else, so the client keeps working with newer versions of Luxe.
+ # Ignore everything else, so the client keeps working with newer versions of luxe.
```
Three rules make a working client: read `ready` first, answer both kinds of prompt, and ignore
@@ -4119,7 +4140,7 @@ client above works with `["luxe", "join"]` in place of `["luxe", "--rpc"]`.
```bash
luxe join --list # sessions running on this machine
luxe join # the one running in this folder
-luxe join 5898d57b # one by the start of its id, or by its Luxe's process id
+luxe join 5898d57b # one by the start of its id, or by its luxe's process id
```
After `ready` comes the session so far: `session_loaded` with the transcript, the latest
@@ -4132,7 +4153,7 @@ A joined client is a guest in someone else's session:
- **Either side can answer a prompt.** The first answer counts, and `prompt_closed` tells everyone
to put the prompt away. The interface says which client answered.
- **It can't switch or end the session.** `new_session` and `load_session` are refused with
- `action_rejected`. `close`, or the end of your stdin, disconnects you and leaves Luxe running.
+ `action_rejected`. `close`, or the end of your stdin, disconnects you and leaves luxe running.
- **GitHub writes are off**, and `ready` says `"github_writes": false`.
- **Keep reading.** A client that falls far behind gets a `notice` and is disconnected. Join
again for a fresh copy of the session.
@@ -4142,7 +4163,7 @@ leaves, by the name it gives in `introduce` if it sends one within a second of c
its process id. A device the bridge introduces as `yours`, paired by you, is listed but not
announced: you know your own phone joined, and a phone reconnects often.
-A running Luxe lists its session in a file in its [runtime folder](environment.md#integrations),
+A running luxe lists its session in a file in its [runtime folder](environment.md#integrations),
`.json`, with the `address` to connect to: a Unix socket, or a named pipe on Windows. Connect
there yourself to skip `luxe join`.
To stop sessions being joined, set [`joinable`](configuration.md#joining-a-session) to `false`.
@@ -4176,12 +4197,12 @@ To stop sessions being joined, set [`joinable`](configuration.md#joining-a-sessi
| `load_file_diff` | `{"request_id": "…", "path": "src/main.rs"}` | One file's change, outside any review; answered by `file_diff_loaded` |
| `list_directory` | `{"request_id": "…", "path": "src"}` | One project folder's entries, without what `.gitignore` ignores unless `"include_ignored": true`, which marks each such entry `"ignored": true`; answered by `directory_listed` |
| `read_project_file` | `{"request_id": "…", "path": "src/main.rs"}` | One project file's text, up to 2 MB; answered by `project_file_read` |
-| `introduce` | `{"name": "iPhone"}` | A joined client's name, which the interface uses when it says who joined or answered. Add `"watch_only": true` as your first line to read without being announced; Luxe then drops everything else you send. `"yours": true` says the person at the terminal paired it, so it's listed without a line in the conversation. Refused from the client that started the session. |
+| `introduce` | `{"name": "iPhone"}` | A joined client's name, which the interface uses when it says who joined or answered. Add `"watch_only": true` as your first line to read without being announced; luxe then drops everything else you send. `"yours": true` says the person at the terminal paired it, so it's listed without a line in the conversation. Refused from the client that started the session. |
| `close` | | Shut down, stopping a turn in flight the way `cancel` does. The stream ends after it. |
### Guided Review
-A review is loaded with `load_guided_review`. Luxe answers with `guided_review_loaded`, which
+A review is loaded with `load_guided_review`. luxe answers with `guided_review_loaded`, which
carries the whole review (its `id`, `revision_id`, files, hunks and chapters), and then always
`review_session_restored`, which carries its comments. Everything below names the review by its
`id`. An action for a review that is no longer the open one is refused.
@@ -4218,7 +4239,7 @@ goes in `edit`, beside `request_id` and `review_id`:
as `review_session_saved` while it runs: the agent's answer to a question, and its report on a fix,
are recorded on the comments. You can keep editing while it works.
-Each comment's `thread` says who said what, by `author`: `reviewer`, `agent`, or `system` for Luxe's
+Each comment's `thread` says who said what, by `author`: `reviewer`, `agent`, or `system` for luxe's
own word. A fix turn that ends with no report on a comment leaves a `system` entry saying why, and
the comment stays `open`.
@@ -4239,7 +4260,7 @@ the comment stays `open`.
| `prepare_thread_answers` | `{"request_id", "review_id"}` | Whether your fixes are pushed yet, on your own pull request; answered by `thread_answers_prepared` |
| `answer_threads` | `{"request_id", "review_id", "answers": [{"finding_id", "reply", "resolve"}]}` | Reply to and resolve its threads, each named by the comment it became. **A GitHub write.** |
-The `open_*` and `post_*` actions confirm a preview by its `request_id`. Luxe keeps what it
+The `open_*` and `post_*` actions confirm a preview by its `request_id`. luxe keeps what it
prepared, so you send back only what a person can change: a title, a description, a verdict.
`post_pull_request_review` is refused if the comments changed after the `session_revision` you drew
the preview from, so what posts is what you showed. A failed open or post can be confirmed again;
@@ -4258,16 +4279,16 @@ These go through your own `gh`, as the interface's `/github` and `/pr` do. `repo
| `read_github_conversation` | `{"number": 142, "pull_request": true}` | Everything said on it; answered by `github_conversation_read` |
| `draft_github_reply` | `{"request_id": "…", "number": 142, "pull_request": true, "target": {…}, "conversation": […]}` | The agent drafts a reply in a turn that changes nothing; answered by `github_reply_drafted`. `target` and `conversation` are what `github_conversation_read` gave you. |
| `check_pull_request_fix` | `{"number": 142}` | Whether it can be checked out here to fix; answered by `pull_request_fix_checked` |
-| `plan_restack` | `{"request_id": "restack:…"}` | What restacking this branch's stack would do; answered by `restack_planned`. Luxe keeps the plan. |
+| `plan_restack` | `{"request_id": "restack:…"}` | What restacking this branch's stack would do; answered by `restack_planned`. luxe keeps the plan. |
| `check_worktree_removal` | | What the session's worktree has not committed; answered by `worktree_removal_checked` |
-| `set_focus` | `{"focused": false}` | Whether your window has focus. Luxe checks GitHub less often while it does not. |
+| `set_focus` | `{"focused": false}` | Whether your window has focus. luxe checks GitHub less often while it does not. |
| `link_pull_request` | `{"number": 142, "url": "…"}` | Record that this session worked on that pull request |
| `post_github_comment` | `{"number": 142, "thread": "PRRT_…", "body": "Fixed."}` | Post a reply in a thread, or a comment without `thread`; answered by `github_comment_posted`. **A GitHub write.** |
| `change_pull_request` | `{"number": 142, "write": {"change": "mark_ready"}}` | `update_branch`, `mark_ready`, `merge`, `enable_auto_merge` or `disable_auto_merge`; answered by `pull_request_changed`. **A GitHub write.** |
| `restack` | `{"request_id": "restack:…"}` | Push the plan `plan_restack` made under that id, exactly as planned; answered by `restacked` or `restack_failed`. **A GitHub write.** |
-You confirm a restack by its id, not by sending the plan back: Luxe pushes the plan it showed you. A
-newer `plan_restack` replaces the older plan, and a plan is pushed once. Before pushing, Luxe checks
+You confirm a restack by its id, not by sending the plan back: luxe pushes the plan it showed you. A
+newer `plan_restack` replaces the older plan, and a plan is pushed once. Before pushing, luxe checks
that every moving branch here is where the plan saw it, checked out where it was, and clean, and
refuses with `restack_failed` otherwise. A push that did not finish is also `restack_failed`, and
its message says to check the branches on GitHub, since some may have moved. If the push succeeds
@@ -4292,7 +4313,7 @@ There are around sixty. These are the ones a front end is built on:
| `kind` | Is |
| --- | --- |
-| `ready` | The handshake: protocol version, Luxe version, session ID, directory |
+| `ready` | The handshake: protocol version, luxe version, session ID, directory |
| `token_delta`, `thinking_delta` | The reply and reasoning as they stream. `block` orders them. |
| `tool_call_composing` | The model has started writing a tool call. Optional; useful for showing a `write` filling in. |
| `tool_started`, `tool_progress`, `tool_completed`, `tool_ended` | A tool call's progress, by `call_id`. `tool_completed` carries its output. |
@@ -4302,7 +4323,7 @@ There are around sixty. These are the ones a front end is built on:
| `question_asked` | **Answer it** with `answer_question` |
| `turn_started` | A turn began, from any client. It comes before the turn's message is saved, and `turn_ended` always follows. A client joining mid-turn gets it as part of catching up. |
| `turn_ended` | The turn finished; `reason` says why |
-| `provider_retrying` | The provider failed before answering, and Luxe is trying again |
+| `provider_retrying` | The provider failed before answering, and luxe is trying again |
| `state_changed` | Model, thinking level, mode, and whether a turn is running |
| `plan_written` | Plan mode wrote `path`. The interface offers the plan's next step when the turn ends. |
| `review_started`, `review_finished` | A requested review began on `model`, and came back with a `verdict` (`continue`, `fix` or `unavailable`) and `findings`. On `fix` the findings also become the agent's next turn. |
@@ -4324,7 +4345,7 @@ Guided Review, as the interface sees it:
| `guided_review_guide_updated` | Better chapters and a summary replaced the first ones, for the same pinned revision |
| `review_session_restored` | The comments and progress that go with the review just loaded. It always follows `guided_review_loaded`, after a restart too. |
| `open_review_shown` | The answer to `show_open_review`, under its `request_id`: the open `review` and its `session`, or neither when no review is open. `reading` says what each chapter asks of the reader: `needs_review` items, and the `reason` when one thing accounts for all of them. |
-| `review_session_saved`, `review_session_save_failed` | A change to them was stored, or refused. Luxe reports its own changes this way too: a turn starting, the agent's report, a review posted. |
+| `review_session_saved`, `review_session_save_failed` | A change to them was stored, or refused. luxe reports its own changes this way too: a turn starting, the agent's report, a review posted. |
| `review_feedback_turn_dispatched`, `review_feedback_turn_completed`, `review_feedback_turn_failed` | Feedback you sent reaching the agent, and how its turn went |
| `review_questions_answered`, `remediation_reported` | The agent's answer to a question, and its report on a fix |
| `review_committed`, `review_commit_failed` | Committing a review's chapters |
@@ -4366,7 +4387,7 @@ list is closed.
- Nothing listens on the network. A [joined](#join-a-running-session) session takes connections
on a local socket or pipe that only your user account can open, and turns away any program the
agent itself started, so the agent can't answer its own prompts.
-- Permission rules are enforced inside Luxe. A client cannot weaken a `deny`: reading `.env` is
+- Permission rules are enforced inside luxe. A client cannot weaken a `deny`: reading `.env` is
refused here as everywhere.
- Prompts go to your client to answer. To approve everything, send `set_permission_auto_approve` or
start with `--auto-approve`. `--on-ask` is refused, since your client can answer.
@@ -4375,27 +4396,27 @@ list is closed.
### GitHub writes
Your client can read GitHub without asking. To let it publish there (post a comment or a review,
-answer threads, open, change, merge or restack a pull request), start Luxe with `--github-writes`.
+answer threads, open, change, merge or restack a pull request), start luxe with `--github-writes`.
Without it, those actions are refused with `action_rejected`, and `ready` says
`"github_writes": false` so you can hide them.
- **Read-only wins.** With `--read-only`, or a read-only mode in your permissions file, the writes
are refused even with the flag, and so is `commit_review_chapters`.
-- **It is a flag, not a setting.** Whoever starts Luxe decides, for that session. No config file
+- **It is a flag, not a setting.** Whoever starts luxe decides, for that session. No config file
turns it on, and a project cannot.
- **It covers your client, not the agent.** The agent's own GitHub writes are permission rules: they
ask first, and the prompt comes to your client. See [Security](security.md#when-you-step-away).
-Luxe is not a sandbox. To isolate it, run it in a [container](security.md#isolation).
+luxe is not a sandbox. To isolate it, run it in a [container](security.md#isolation).
@@@ page: security
---
title: "Security and permissions"
-summary: "What the agent may do, the rules and prompts that decide it, secrets, and what Luxe sends over the network."
+summary: "What the agent may do, the rules and prompts that decide it, secrets, and what luxe sends over the network."
order: 1
group: "Reference"
---
-The agent reads and writes files and runs commands on your machine. Luxe lets ordinary work through
+The agent reads and writes files and runs commands on your machine. luxe lets ordinary work through
without asking, keeps a few guardrails on at all times, and gives you rules to tighten the rest.
For real isolation, run it in a [container](#isolation).
@@ -4495,7 +4516,7 @@ open. They don't sandbox a language server, which can index other files and runs
code.
A project's `.luxe/permissions.json` can only make your rules stricter, never looser, unless you
-start Luxe with `--trust-project`.
+start luxe with `--trust-project`.
## The permission prompt
@@ -4541,11 +4562,11 @@ which then sees the conversation and can answer its prompts. That makes who may
question:
- **Only your user account.** The socket sits in a folder only you can open, only you can use it,
- and Luxe checks the account of whatever connects. On Windows the named pipe admits your account
+ and luxe checks the account of whatever connects. On Windows the named pipe admits your account
alone, and refuses remote clients.
Nothing listens on the network.
- **Not the agent.** A program the agent started, at any depth, is turned away, so it can't
- approve its own prompts. Luxe checks the joining process's ancestry. It can't see past a program
+ approve its own prompts. luxe checks the joining process's ancestry. It can't see past a program
that detaches from its parent, such as one started with `setsid` and left running after its
parent exits, so treat this as a tripwire, not a wall. A command that would do that should be
something your rules ask about.
@@ -4556,19 +4577,19 @@ Anything running as you can already read your sessions on disk and your credenti
adds a way to steer, not a way to read. To turn it off, set
[`joinable`](configuration.md#joining-a-session) to `false`.
-## Luxe remote
+## luxe remote
-[`luxe remote`](remote.md) is the one part of Luxe that answers network requests, and it is off
+[`luxe remote`](remote.md) is the one part of luxe that answers network requests, and it is off
until you run it. It puts your sessions in reach of your phone, so it takes two things to get in:
- **Your Tailscale login.** The bridge listens on this machine only (`127.0.0.1`) and refuses to
listen anywhere else. Tailscale Serve carries it to your tailnet over HTTPS and adds the login of
whoever is asking, replacing any the request brought. A request without one didn't come through
- Serve and is refused, and so is a login other than yours or one you `--allow`. Luxe never uses
+ Serve and is refused, and so is a login other than yours or one you `--allow`. luxe never uses
Tailscale Funnel, which would publish it to the internet.
- **A paired device.** Any program on this machine can reach a local port and claim a login, so
every session also needs the credential a device gets when it pairs. An invitation works once,
- within ten minutes. Luxe keeps only a hash of each invitation and credential, in
+ within ten minutes. luxe keeps only a hash of each invitation and credential, in
`~/.config/luxe/remote.json`. The device keeps its credential in a cookie scripts can't read,
sent only over HTTPS and never with a request from another site; a request a page on another
site makes is refused too.
@@ -4582,28 +4603,28 @@ Notifications are off until you turn them on in the app, one device at a time, a
answer anything: a notification says what is waiting and where, such as "Approve bash" over the
folder's name, never the command, and answering takes the app. While any device has them on, the
bridge joins each running session to watch for prompts. It joins watch-only: the terminal doesn't
-announce it, and Luxe drops anything it sends, so a client nobody sees can't act. Each
+announce it, and luxe drops anything it sends, so a client nobody sees can't act. Each
notification is sealed so only that device can read it (RFC 8291) and posted to the push service
the device's browser picked; the push service sees when, not what. The bridge posts only to
Apple's, Google's, Mozilla's and Microsoft's push services, so a paired device can't aim it at
anything else on your network. It signs with a key of its own, kept in `remote.json`, and revoking
a device ends its notifications.
-The bridge starts Luxe only in a folder a device asks for, as you, in a herdr pane when herdr is
-already running and as its own `luxe --rpc` otherwise. A packager can build Luxe without the
+The bridge starts luxe only in a folder a device asks for, as you, in a herdr pane when herdr is
+already running and as its own `luxe --rpc` otherwise. A packager can build luxe without the
`remote` feature, which leaves out the bridge entirely.
-A session is open in one Luxe at a time. A second `luxe --resume` of a session the interface
+A session is open in one luxe at a time. A second `luxe --resume` of a session the interface
already has, or a resume from the session list, is refused and says which process has it. The lock
is the operating system's, so a crash never leaves a session stuck.
## Secrets
-- Credentials are stored in `~/.config/luxe/auth.json`, readable only by you. Luxe warns if
+- Credentials are stored in `~/.config/luxe/auth.json`, readable only by you. luxe warns if
`auth.json` or `models.json` can be read by others.
- A key is found in this order: `--api-key` or `LUXE_API_KEY`, then what you signed in with, then
the provider's own variable, then a provider block in `auth.json` or `models.json`. For a provider
- Luxe does not know, the block comes before `OPENAI_API_KEY`. The order is the same for the
+ luxe does not know, the block comes before `OPENAI_API_KEY`. The order is the same for the
provider a session starts on and for one you switch to in `/model`.
- A key can be `$VAR` or `!command` rather than the value, so it never sits in a file. A command
runs when a key is needed, so treat these files like your shell's startup file. It runs from your
@@ -4611,7 +4632,7 @@ is the operating system's, so a crash never leaves a session stuck.
needs a graphical prompt.
- Anything that looks like a secret (API keys, tokens, private keys, passwords in URLs) is removed
from tool output before the model, the transcript, hooks or the log see it. So is every key and
- token Luxe itself has used or loaded, whatever it looks like, including `$VAR` and `!command`
+ token luxe itself has used or loaded, whatever it looks like, including `$VAR` and `!command`
keys and MCP tokens, as long as it is at least 8 characters.
- MCP sign-in tokens are kept in `~/.local/state/luxe/mcp-auth.json`, outside your config folder,
so a dotfiles repository never carries them.
@@ -4665,23 +4686,23 @@ Opening a repository should not run its code. So, without `--trust-project`:
[Tools](tools.md#a-server-the-project-installed).
Its `AGENTS.md` and skills, including the ones it commits in `.agents/skills/`, still reach the
-model, because your own project's are the point of the feature. Luxe removes text a reviewer
+model, because your own project's are the point of the feature. luxe removes text a reviewer
cannot see, comments and invisible Unicode, and leaves out a file it cannot finish checking; see
-[What Luxe removes](project-context.md#what-luxe-removes). Plain text is still instructions, so read
+[What luxe removes](project-context.md#what-luxe-removes). Plain text is still instructions, so read
an unfamiliar repository's `AGENTS.md` as you would its `Makefile`.
Text from GitHub, such as issues and comments, is filtered the same way and given to the agent as
someone else's words, not as instructions. So are MCP servers' descriptions and results.
A language server can run project code itself: rust-analyzer runs `build.rs`. Treat opening a
-repository in Luxe like building it.
+repository in luxe like building it.
-A repository the agent fetches with `web_fetch` is different: you never chose to open it. Luxe
+A repository the agent fetches with `web_fetch` is different: you never chose to open it. luxe
clones it into `.luxe/gh-cache/` with symbolic links flattened to plain files, and starts no
language server or formatter there; see [Tools](tools.md#github-repositories).
### Isolation
-Rules are a policy, not a sandbox. For hard isolation, run Luxe in a container that mounts only the
+Rules are a policy, not a sandbox. For hard isolation, run luxe in a container that mounts only the
project:
```bash
@@ -4692,9 +4713,9 @@ docker run --rm -it -v "$PWD":/work -w /work -e ANTHROPIC_API_KEY ` |
@@ -4782,7 +4803,7 @@ threshold. A price you set in `models.json` replaces the catalog's, tiers and al
Anthropic keeps a cached prompt for five minutes after it was last read. A turn that waits longer
than that, on a slow command, a permission prompt or a question, loses the cache, and the next
request pays to write the whole conversation again: about $2 on a 400k-token Opus session. So while
-a turn runs, Luxe re-sends its last request with a one-token reply just before the five minutes are
+a turn runs, luxe re-sends its last request with a one-token reply just before the five minutes are
up. That costs a cache read, a tenth of the write it saves.
`agent.cache_warming` decides when:
@@ -4794,7 +4815,7 @@ up. That costs a cache read, a tenth of the write it saves.
| `off` | Never |
A refresh goes out only when it is expected to save at least $0.05, so a short conversation is
-never refreshed. Between turns Luxe cannot know you will reply, so `idle` refreshes only a large
+never refreshed. Between turns luxe cannot know you will reply, so `idle` refreshes only a large
conversation. On real sessions it lost money overall, which is why it is not the default. Refreshes
are never sent on a subscription plan, or for an Anthropic model using a fixed thinking budget,
whose cache would change on replay. Their cost shows under **cache warming** in `/cost`.
@@ -4824,7 +4845,7 @@ more on every turn instead of on each pause.
| tool schemas | The tools the agent can call | `tools.disabled`; see [Tools](tools.md) |
| instructions | Your `AGENTS.md` files | A shorter file; see [Project context](project-context.md) |
| skills | The list of skills the agent can load | Fewer, or shorter descriptions |
-| harness | Luxe's own instructions | Nothing; this is fixed |
+| harness | luxe's own instructions | Nothing; this is fixed |
| appended | One-time context added to a resumed session, such as what changed while you were away | `/new` |
The largest single items are named under a row, which is usually how you find a `read` or `bash`
@@ -4832,7 +4853,7 @@ call that should have been narrower. The total is what the provider reported for
## Compaction
-When a conversation fills 90% of the usable window, Luxe summarizes it and carries on, keeping the
+When a conversation fills 90% of the usable window, luxe summarizes it and carries on, keeping the
most recent part word for word. The summary records the goal, constraints, progress, decisions,
next steps and the files the session touched. You see a line like
`Compacted the conversation — 82k → 21k of context.`
@@ -4841,7 +4862,7 @@ next steps and the files the session touched. You see a line like
conversation that already fits in what `keep_recent` keeps has nothing to summarize, and `/compact`
says so.
-Before summarizing, from 80% full, Luxe first replaces old tool output in the request with a short
+Before summarizing, from 80% full, luxe first replaces old tool output in the request with a short
placeholder. The session file keeps everything; only what is sent changes.
To change when and how it happens:
@@ -4869,7 +4890,7 @@ To change when and how it happens:
| `summarizer` | the session's model | The model that writes the summary, as `provider/id`, on a provider you have set up. A smaller, cheaper model is usually enough, and `/cost` prices its spend as it does any other model's. |
This works in a project's `.luxe/config.json` too, and so does `agent.cache_warming`, which a project
-can turn down but not up. The compaction settings in `/settings` apply the next time Luxe starts.
+can turn down but not up. The compaction settings in `/settings` apply the next time luxe starts.
## Long command output
@@ -4906,7 +4927,7 @@ One folder per skill, with a `SKILL.md` in it:
| `.luxe/skills//` | This project. Git sees it too, since `.luxe/` keeps `skills/` out of what it ignores. |
`.agents/skills/` is where Codex, OpenCode, Pi, GitHub Copilot and the `npx skills` installer put
-skills, so one installed for any of them works here too. Luxe reads a repository's `.agents/skills/`
+skills, so one installed for any of them works here too. luxe reads a repository's `.agents/skills/`
from the top of the repository down to the project. When two skills share a name, whatever its case,
the one lower in the table wins, so a project's skill replaces your own, and between nested
`.agents/skills/` folders the deeper one wins. A skill reached twice through a symbolic link counts
@@ -4939,8 +4960,8 @@ with a note telling the agent to read the rest of the file.
The agent loads a skill when its description fits the task. To ask for one yourself, mention it:
`@`, `tab` to **Skills**, and `enter`, as in *split this into a stack with @[skill:gh-stack]*.
-**Project context** lists the skills Luxe found: type `skills` on the start screen and open it. A
-`SKILL.md` placed directly in `skills/`, without its own folder, is skipped, and Luxe names it at
+**Project context** lists the skills luxe found: type `skills` on the start screen and open it. A
+`SKILL.md` placed directly in `skills/`, without its own folder, is skipped, and luxe names it at
startup. The list in the agent's instructions is read with a session's first message, and again
when you resume one. The `@` picker and the `skill` tool look again every time, so a skill added
mid-session can be mentioned or loaded at once.
@@ -4961,7 +4982,7 @@ itself, one level too high.
A skill's text reaches the model as instructions, so read a skill before installing it, as you
would a script. The same goes for a repository you clone: its `.agents/skills/` and
-`.luxe/skills/` reach the model without `--trust-project`. Luxe removes comments and invisible
+`.luxe/skills/` reach the model without `--trust-project`. luxe removes comments and invisible
characters from skill files, as it does from [`AGENTS.md`](project-context.md#what-luxe-removes),
but without a notice.
@@@ page: tools
@@ -4969,7 +4990,7 @@ but without a notice.
title: "Tools"
summary: "What the agent can use: files, the shell, tests, Git, code intelligence, a debugger and the web."
order: 10
-group: "Using Luxe"
+group: "Using luxe"
---
## What the agent can use
@@ -5017,7 +5038,7 @@ the file:
{ "tools": { "disabled": ["web_search", "screenshot"] } }
```
-A change applies the next time Luxe starts, since the tools the agent is offered are fixed for a
+A change applies the next time luxe starts, since the tools the agent is offered are fixed for a
run.
For one session, `luxe --no-tool web_search` turns a tool off and `luxe --with-tool web_search`
@@ -5038,7 +5059,7 @@ their config, in the file's directory or one above it:
| TypeScript, JavaScript | biome | `biome.json` or `biome.jsonc` |
| TOML | taplo | `taplo.toml` or `.taplo.toml` |
-A formatter that is not installed is skipped. When Luxe pipes a file through one (taplo), an empty
+A formatter that is not installed is skipped. When luxe pipes a file through one (taplo), an empty
answer, or a file that changed again while it ran, leaves the file as it was. To turn formatting
off:
@@ -5059,12 +5080,12 @@ reported as having had no effect; one stopped partway says so, since it may have
A command that starts something and leaves it running, such as a dev server, returns when the
command itself ends. The server keeps running.
-Tools work from the directory Luxe was started in. A `cd` inside one `bash` call does not carry over
-to the next, so start Luxe in the project you want to work on.
+Tools work from the directory luxe was started in. A `cd` inside one `bash` call does not carry over
+to the next, so start luxe in the project you want to work on.
## Code intelligence
-After every edit, Luxe asks the file's language server for errors and warnings and adds the ones
+After every edit, luxe asks the file's language server for errors and warnings and adds the ones
the edit introduced to the result the agent sees, so it can fix what it broke before moving on.
Problems that were already in the file are counted, not listed. The result also names changed code
that no test covers. Set `LUXE_LSP_ON_EDIT=0` to turn both off.
@@ -5075,7 +5096,7 @@ while the change is fresh. A file that was already long says nothing more. Funct
in Rust, Python, Go, TypeScript and JavaScript; files in any language. Set `LUXE_SIZE_CHECK=0` to turn this off, for a project whose own
conventions differ.
-Luxe launches these language servers when they are installed:
+luxe launches these language servers when they are installed:
| Language | Server | Install |
| --- | --- | --- |
@@ -5096,13 +5117,13 @@ anywhere else, or inside a [fetched repository](#github-repositories), is refuse
every file is written or none is, and the result says which edits didn't fit their file and which
file operations, such as creating or renaming one, it left undone.
-`/lsp` shows which servers are running, and `luxe doctor` starts each one to check it works. Luxe
+`/lsp` shows which servers are running, and `luxe doctor` starts each one to check it works. luxe
never installs a server itself. When one is missing, the result tells the agent the command that
installs it, which runs through `bash` under your rules like any other.
### A server the project installed
-Luxe prefers the copy a project installed for itself, in its `node_modules/.bin` or virtual
+luxe prefers the copy a project installed for itself, in its `node_modules/.bin` or virtual
environment, since that version matches the project. The same goes for formatters such as `ruff`
and `biome`, and for MCP servers a project declares. Runs with nobody watching, such as
`--headless` and the loop, use only what you installed yourself unless you pass `--trust-project`.
@@ -5132,14 +5153,14 @@ use a container rather than relying on this setting.
}
```
-An entry for a language Luxe already knows changes only what it names, so
+An entry for a language luxe already knows changes only what it names, so
`{ "rust": { "command": "rust-analyzer-nightly" } }` swaps the binary and keeps the rest.
`alternatives` lists other servers to try, in order, when the first will not start.
## Debugger
The `debug` tool drives a real debugger, so the agent can set breakpoints, step through code and
-inspect variables rather than add print statements. After three failing test runs in a row, Luxe
+inspect variables rather than add print statements. After three failing test runs in a row, luxe
suggests it. `LUXE_DEBUG_NUDGE=0` turns that suggestion off.
| Language | Adapter |
@@ -5163,9 +5184,9 @@ with `lldb-dap`:
]
```
-`transport` is `stdio` (the default) or `tcp_dial_in`, for an adapter that connects back to Luxe:
+`transport` is `stdio` (the default) or `tcp_dial_in`, for an adapter that connects back to luxe:
put `${DAP_ADDR}` in its `command` where it takes the address, as Delve's
-`dlv dap --client-addr ${DAP_ADDR}` does, and Luxe replaces it with `host:port`.
+`dlv dap --client-addr ${DAP_ADDR}` does, and luxe replaces it with `host:port`.
`addr_placeholder` names a different placeholder.
A breakpoint the adapter could not set is listed under the launch, with the adapter's reason. An
@@ -5192,7 +5213,7 @@ server works. Other machines on your network, cloud metadata addresses and `file
including through an IPv6 address that carries one of them. A page elsewhere can't redirect a fetch
to your machine, and a fetch reads at most 10 MB. `screenshot` loads only the host you asked for, so
a redirect, frame or image from any other host fails to load, and it runs the browser on a fresh
-profile, sandboxed unless Luxe runs as root or in a container. `LUXE_WEB_ALLOW_PRIVATE=1` lifts the
+profile, sandboxed unless luxe runs as root or in a container. `LUXE_WEB_ALLOW_PRIVATE=1` lifts the
address checks, allows `file://` in `screenshot`, and lets it load other hosts. A fetch gives up
after 30 seconds, or `LUXE_WEB_TIMEOUT_SECS`, within the call's 60-second limit.
@@ -5206,7 +5227,7 @@ of reading the page, so the agent can search the real files. The clone is shallo
only, and goes into `.luxe/gh-cache/owner/repo` in your project, which git ignores. Reading it never
asks permission, because it is inside the project.
-A repository you did not write sits inside one you did, so Luxe keeps it inert:
+A repository you did not write sits inside one you did, so luxe keeps it inert:
- Symbolic links are checked out as plain files holding the link's target. A link cannot point the
agent at your `.env`, or anywhere else outside the clone.
@@ -5230,7 +5251,7 @@ other page from the web. To read repositories as web pages instead:
@@@ page: troubleshooting
---
title: "Troubleshooting"
-summary: "Common problems, where Luxe logs, and what to include in a bug report."
+summary: "Common problems, where luxe logs, and what to include in a bug report."
order: 4
group: "Reference"
---
@@ -5241,23 +5262,23 @@ group: "Reference"
luxe doctor
```
-It lists the tools, language servers and debug adapters Luxe found, starts each language server to
+It lists the tools, language servers and debug adapters luxe found, starts each language server to
check that it works, and shows how to install or fix what does not.
## Common problems
| Problem | Fix |
| --- | --- |
-| **No model selected**, or `--headless` refuses to start | Choose one with `/model`, or pass `--model provider/id`. Luxe never picks one for you. |
+| **No model selected**, or `--headless` refuses to start | Choose one with `/model`, or pass `--model provider/id`. luxe never picks one for you. |
| The model picker is empty | No provider is set up. Open **Configuration** (`c`), or set a provider's key in the environment. |
| A provider keeps failing with `retrying (3/3)` | It is overloaded or unreachable. See [When a provider stalls](providers.md#when-a-provider-stalls) for the time limits. |
-| A setting does nothing | Luxe ignores keys it does not know, so check the spelling. `luxe init` and `/settings` write a `config.schema.json` your editor can check against. A project's file cannot change some settings; see [Configuration](configuration.md#what-a-project-cannot-change). |
+| A setting does nothing | luxe ignores keys it does not know, so check the spelling. `luxe init` and `/settings` write a `config.schema.json` your editor can check against. A project's file cannot change some settings; see [Configuration](configuration.md#what-a-project-cannot-change). |
| Icons show as boxes, or sit flush against their words | See [Fonts](install.md#fonts), and the `glyphs` setting in [Appearance](appearance.md#icons-and-fonts). |
| Your terminal's copy key does nothing | The terminal keeps it. Hold `shift` while dragging, or let the key pass through; see [Mouse and clipboard](interface.md#mouse-and-clipboard). |
| Images do not appear | Your terminal may not draw them. `/images` says why; see [Images](appearance.md#images). |
| A language server is not used | `luxe doctor` says whether it starts. A server a project installed is used only when you are watching; see [Tools](tools.md#a-server-the-project-installed). |
| A project's hooks or MCP servers do not run | They need `--trust-project`, or an interactive session for MCP servers. See [Security](security.md#untrusted-repositories). |
-| Luxe opened the wrong folder as the project | Inside a repository, a `.luxe/` between you and its top scopes the project there. Remove it, or start Luxe somewhere else. See [Where the project begins](project-context.md#where-the-project-begins). |
+| luxe opened the wrong folder as the project | Inside a repository, a `.luxe/` between you and its top scopes the project there. Remove it, or start luxe somewhere else. See [Where the project begins](project-context.md#where-the-project-begins). |
| `/pr` or `/github` says GitHub is unavailable | Install the GitHub CLI and sign in: `gh auth login`. A private repository also needs `gh auth setup-git`. |
| A key binding does not work | Your terminal may not send that key. Rebind it; see [Commands and keys](commands.md#rebind-keys). |
| On a Mac, some `alt` keys type a character such as `ø` or `π` | Your terminal treats Option as a key that types characters. Set it to act as Alt; see [Alt keys on a Mac](commands.md#alt-keys-on-a-mac). |
@@ -5268,9 +5289,9 @@ A session logs to `.luxe/logs/luxe.log` in the project. Commands that only repor
`luxe auth`, `luxe doctor` and `luxe models`, log to `logs/luxe.log` in the state directory:
`~/.local/state/luxe/` (`%LOCALAPPDATA%\luxe\` on Windows), or wherever `LUXE_STATE_DIR` or
`XDG_STATE_HOME` puts it. On macOS and Linux, a log is readable only by you. One past 5 MB is moved
-to `luxe.log.1` the next time Luxe starts, replacing the one before.
+to `luxe.log.1` the next time luxe starts, replacing the one before.
-The log shows Luxe's own activity at `info` and everything else at `warn`. For more:
+The log shows luxe's own activity at `info` and everything else at `warn`. For more:
```bash
LUXE_LOG=debug luxe # everything
@@ -5279,7 +5300,7 @@ LUXE_LOG=luxe_core::hooks=debug luxe # one part, in detail
## See what is sent to the model
-`LUXE_WIRE_LOG` records every request Luxe sends and every reply, with credentials removed.
+`LUXE_WIRE_LOG` records every request luxe sends and every reply, with credentials removed.
`luxe wire` summarizes it: what fills each request, whether the cached part stays the same between
turns, and whether anything is sent twice.
@@ -5300,7 +5321,7 @@ Open an issue at [github.com/evanwhittaker/luxe](https://github.com/evanwhittake
- The relevant part of `.luxe/logs/luxe.log`
If the interface itself draws wrongly, such as stray characters or a half-drawn screen, record the
-raw output, which shows what Luxe actually wrote:
+raw output, which shows what luxe actually wrote:
```bash
script -f /tmp/luxe.tty -c luxe # replay with `cat` or `scriptreplay`
diff --git a/crates/luxe-cli/src/guide_check.rs b/crates/luxe-cli/src/guide_check.rs
index d0fc87a..a2434b6 100644
--- a/crates/luxe-cli/src/guide_check.rs
+++ b/crates/luxe-cli/src/guide_check.rs
@@ -1,9 +1,9 @@
//! Guide check — the user guide's config examples, flags and slash commands, held against the code.
//!
-//! `docs/guide/` is what a user copies from and what an agent follows. Luxe ignores a config key it
-//! does not know, on purpose, so an old config loads on a new Luxe; the cost is that a misspelled
+//! `docs/guide/` is what a user copies from and what an agent follows. luxe ignores a config key it
+//! does not know, on purpose, so an old config loads on a new luxe; the cost is that a misspelled
//! or invented key in the docs loads too, and does nothing. These tests load every example the
-//! guide names a file for as that file, and fail on anything Luxe would ignore.
+//! guide names a file for as that file, and fail on anything luxe would ignore.
//!
//! A code block opts in by naming its file in the fence, ` ```json title="config.json" `, which is
//! also what the site shows above the block. `scripts/docs-check.py` covers links and voice.
@@ -76,7 +76,7 @@ fn blocks(page: &str, text: &str) -> Vec {
blocks
}
-/// The keys Luxe would ignore reading `text` as `T`, as paths like `dock.mode`.
+/// The keys luxe would ignore reading `text` as `T`, as paths like `dock.mode`.
fn ignored(text: &str) -> Result, String> {
let value: serde_json::Value = serde_json::from_str(text).map_err(|error| error.to_string())?;
let mut paths = Vec::new();
@@ -85,7 +85,7 @@ fn ignored(text: &str) -> Result, String> {
Ok(paths)
}
-/// The keys Luxe would ignore in an `auth.json`: every top-level key is a provider's credential,
+/// The keys luxe would ignore in an `auth.json`: every top-level key is a provider's credential,
/// except the configuration keys it shares with `models.json`.
fn auth_file(text: &str) -> Result, String> {
const CONFIGURATION: &[&str] =
@@ -153,7 +153,7 @@ fn check_file(title: &str, text: &str) -> Result<(), String> {
if unknown.is_empty() {
Ok(())
} else {
- Err(format!("Luxe would ignore {}", unknown.join(", ")))
+ Err(format!("luxe would ignore {}", unknown.join(", ")))
}
}
@@ -174,7 +174,7 @@ fn every_named_config_example_loads() {
}
assert!(
failures.is_empty(),
- "config examples Luxe would not read as written:\n{}",
+ "config examples luxe would not read as written:\n{}",
failures.join("\n")
);
}
@@ -235,7 +235,7 @@ fn shell_words(line: &str) -> Vec {
words
}
-/// What is wrong with a `luxe …` command line, if anything: a flag or subcommand Luxe lacks.
+/// What is wrong with a `luxe …` command line, if anything: a flag or subcommand luxe lacks.
fn check_command_line(root: &clap::Command, line: &str) -> Result<(), String> {
let mut path: Vec<&clap::Command> = vec![root];
let words = shell_words(line);
@@ -286,14 +286,14 @@ fn every_command_line_in_the_guide_exists() {
}
}
}
- assert!(failures.is_empty(), "command lines Luxe would refuse:\n{}", failures.join("\n"));
+ assert!(failures.is_empty(), "command lines luxe would refuse:\n{}", failures.join("\n"));
}
/// Paths that look like a slash command in a code span.
const NOT_COMMANDS: &[&str] =
&["/tmp", "/etc", "/usr", "/dev", "/home", "/var", "/opt", "/bin", "/proc"];
-/// Every slash command the guide names is one Luxe has.
+/// Every slash command the guide names is one luxe has.
#[test]
fn every_slash_command_in_the_guide_exists() {
let registry = luxe_tui::commands::CommandRegistry::all();
@@ -329,10 +329,10 @@ fn every_slash_command_in_the_guide_exists() {
}
}
}
- assert!(failures.is_empty(), "slash commands Luxe does not have:\n{}", failures.join("\n"));
+ assert!(failures.is_empty(), "slash commands luxe does not have:\n{}", failures.join("\n"));
}
-/// The command reference lists every slash command Luxe has, and every key it binds by default, so a
+/// The command reference lists every slash command luxe has, and every key it binds by default, so a
/// new command or binding cannot ship undocumented.
#[test]
fn the_command_reference_is_complete() {
@@ -350,7 +350,7 @@ fn the_command_reference_is_complete() {
assert!(missing.is_empty(), "commands.md does not mention:\n{}", missing.join("\n"));
}
-/// The configuration reference names every key `config.json` takes, from the schema Luxe writes
+/// The configuration reference names every key `config.json` takes, from the schema luxe writes
/// beside it, so a new setting cannot ship undocumented.
#[test]
fn the_configuration_reference_is_complete() {
diff --git a/crates/luxe-cli/src/logging.rs b/crates/luxe-cli/src/logging.rs
index ca266df..aa0924a 100644
--- a/crates/luxe-cli/src/logging.rs
+++ b/crates/luxe-cli/src/logging.rs
@@ -11,7 +11,7 @@
//! ranks a directory holding one above a nearer checkout, so a stray left in `/tmp` re-roots every
//! repository beneath it.
//!
-//! Level is controlled by `LUXE_LOG` (or `RUST_LOG`). The default is `info` for Luxe's own crates
+//! Level is controlled by `LUXE_LOG` (or `RUST_LOG`). The default is `info` for luxe's own crates
//! and `warn` for everything else — a normal run leaves a readable account of what the agent did,
//! without `hyper` and `rustls` narrating every socket. Init is best-effort: logging must never
//! crash or block the app.
@@ -54,7 +54,7 @@ impl<'a> MakeWriter<'a> for RedactingFile {
}
}
-/// Luxe's own crates, which is what the default filter turns up.
+/// luxe's own crates, which is what the default filter turns up.
///
/// Spelled out rather than matched by prefix because `EnvFilter` has no glob for a target prefix, and
/// a filter that quietly matched nothing would look exactly like a quiet run.
@@ -69,7 +69,7 @@ const LUXE_TARGETS: [&str; 8] = [
"luxe_tools",
];
-/// The default filter: Luxe at `info`, everything else at `warn`.
+/// The default filter: luxe at `info`, everything else at `warn`.
///
/// Two things have to be true at once. A normal run should leave an account someone can read — which
/// `warn` did not, and which is why routine loop behavior had drifted up to `warn!` just to be
@@ -167,22 +167,22 @@ fn open_log(dir: &Path, rotate_at: u64) -> io::Result<(File, Option)>
#[cfg(unix)]
std::os::unix::fs::OpenOptionsExt::mode(&mut options, 0o600);
let file = options.open(&path)?;
- // a log an older Luxe created at the umask's default is tightened too
+ // a log an older luxe created at the umask's default is tightened too
#[cfg(unix)]
file.set_permissions(std::os::unix::fs::PermissionsExt::from_mode(0o600))?;
Ok((file, rotated))
}
-/// Drop a self-ignoring `.gitignore` into `.luxe/`, and upgrade one Luxe wrote itself.
+/// Drop a self-ignoring `.gitignore` into `.luxe/`, and upgrade one luxe wrote itself.
///
-/// The directory ignores itself (the trick pytest/ruff/venv use), so running Luxe inside any repo
+/// The directory ignores itself (the trick pytest/ruff/venv use), so running luxe inside any repo
/// needs no root-`.gitignore` edit or global git config. The rule is an *allowlist*: everything is
/// machine-local by default (`*` — sessions, logs, caches, review memory, per-user UI prefs), with
/// explicit exceptions for the handful of project-config files a team may want to version
/// (`permissions.json`, `mcp.json`, `hooks.json`, `skills/`). Ignoring by default is the safe
-/// direction: a new artifact kind Luxe writes later stays out of git until someone opts it in.
+/// direction: a new artifact kind luxe writes later stays out of git until someone opts it in.
///
-/// A file Luxe did not author is never touched, so a user's own curation survives.
+/// A file luxe did not author is never touched, so a user's own curation survives.
fn ensure_self_ignored(luxe_dir: &Path) {
let ignore = luxe_dir.join(".gitignore");
match std::fs::read_to_string(&ignore) {
@@ -219,7 +219,7 @@ const SELF_IGNORE_BODY: &str = "\
# and luxe keeps it out of its own reviews and changes lists for the same reason.
*
-# Exceptions: project config you may want to version and share. Luxe rewrites
+# Exceptions: project config you may want to version and share. luxe rewrites
# this file while its first line says it made it; to choose for yourself, replace
# the file, first line included, with your own.
!permissions.json
diff --git a/crates/luxe-cli/src/logging/tests.rs b/crates/luxe-cli/src/logging/tests.rs
index 21c4887..930e3c0 100644
--- a/crates/luxe-cli/src/logging/tests.rs
+++ b/crates/luxe-cli/src/logging/tests.rs
@@ -1,9 +1,9 @@
use super::*;
use std::io::Write;
-/// Luxe's own crates are visible at `info`, and dependencies are not.
+/// luxe's own crates are visible at `info`, and dependencies are not.
///
-/// The break this catches is the one that shipped: with a bare `warn` default, nothing Luxe
+/// The break this catches is the one that shipped: with a bare `warn` default, nothing luxe
/// logged at `info` was ever seen, so routine loop behavior drifted up to `warn!` just to be
/// visible — an ordinary steer arrived labeled as a warning. Widening it to `info` for
/// everything instead would swap that for a firehose, because `hyper` and `rustls` narrate every
@@ -46,7 +46,7 @@ fn the_default_filter_shows_luxe_at_info_and_nothing_else() {
let seen = seen.lock().expect("not poisoned").clone();
assert!(
seen.contains(&"luxe_core@INFO".to_string()),
- "Luxe's own routine logging must be visible: {seen:?}"
+ "luxe's own routine logging must be visible: {seen:?}"
);
assert!(!seen.contains(&"luxe_core@DEBUG".to_string()), "debug stays opt-in: {seen:?}");
assert!(
@@ -153,7 +153,7 @@ fn self_ignore_hides_the_whole_directory_and_never_clobbers() {
}
/// A log holds what the agent read and ran, so a new one is readable by its owner only, and one an
-/// older Luxe left at the umask's default is tightened when it is next opened.
+/// older luxe left at the umask's default is tightened when it is next opened.
#[cfg(unix)]
#[test]
fn a_log_file_is_owner_only() {
diff --git a/crates/luxe-cli/src/loop_run.rs b/crates/luxe-cli/src/loop_run.rs
index fc71dbf..218df60 100644
--- a/crates/luxe-cli/src/loop_run.rs
+++ b/crates/luxe-cli/src/loop_run.rs
@@ -179,7 +179,7 @@ mod tests {
external_work: None,
review_context_files: Vec::new(),
review_exclude: Vec::new(),
- remember_runtime_state: false,
+ remember_choices_in: None,
cache_warming: luxe_core::cache_warming::CacheWarming::Off,
review_insight_provider: None,
github: None,
@@ -197,7 +197,7 @@ mod tests {
event_rx,
running_rx,
"Continue building.".into(),
- crate::loop_view::LoopView::new(true),
+ crate::loop_view::LoopView::new(true, 0),
)
.await
});
diff --git a/crates/luxe-cli/src/loop_view.rs b/crates/luxe-cli/src/loop_view.rs
index db6e5bc..c4abd79 100644
--- a/crates/luxe-cli/src/loop_view.rs
+++ b/crates/luxe-cli/src/loop_view.rs
@@ -14,12 +14,20 @@
//! Timed from the run's own start rather than the wall clock. "Twenty minutes in and still on
//! phase 0" is the shape of the question being asked, and the absolute time is in the log for
//! anything forensic.
+//!
+//! Each line is led by guides for the phase and iteration it belongs to ([`paint`]), and a tool
+//! call's line takes the color of its result once it has one ([`screen`]).
-use crate::output::outln;
+mod paint;
+mod screen;
+
+use paint::{CallState, Depth, Edge, Paint};
+use screen::{Printed, Screen};
use std::collections::HashMap;
use std::time::Instant;
use luxe_core::event::Event;
+use luxe_core::hooks::self_continuation::parse_iteration_notice;
use luxe_core::message::{ContentBlock, Role};
/// Longest assistant line echoed into the view.
@@ -43,16 +51,47 @@ pub(crate) struct LoopView {
tools_this_turn: usize,
/// When the turn now running began.
turn_started: Option,
+ /// The phase the run is on, once known
+ phase: Option,
+ /// Whether an iteration's rail is open
+ in_iteration: bool,
+ /// The phase the run started on, until its first turn opens it: no iteration notice does
+ start_phase: Option,
+ /// Calls that have started and not ended, with their line and what it said
+ running: HashMap,
+ paint: Paint,
+ screen: Screen,
+}
+
+/// A started call's line, kept to redraw it with the result.
+struct RunningCall {
+ printed: Printed,
+ /// When in the run it was printed, so the redraw keeps its time
+ at: std::time::Duration,
+ name: String,
+ label: String,
+ depth: Depth,
}
impl LoopView {
- pub(crate) fn new(quiet: bool) -> Self {
+ /// A view for a run starting on `phase`, in color on a terminal.
+ pub(crate) fn new(quiet: bool, phase: u32) -> Self {
+ Self::with(quiet, Some(phase), Paint::detect(), Screen::detect())
+ }
+
+ fn with(quiet: bool, start_phase: Option, paint: Paint, screen: Screen) -> Self {
Self {
quiet,
started: Instant::now(),
labels: HashMap::new(),
tools_this_turn: 0,
turn_started: None,
+ phase: None,
+ in_iteration: false,
+ start_phase,
+ running: HashMap::new(),
+ paint,
+ screen,
}
}
@@ -64,9 +103,14 @@ impl LoopView {
match event {
// A hook talking: the phase moved, the gate refused, the run is blocked. These are the
// lines the whole view exists for.
- Event::Notice(message) => self.line("", message),
+ Event::Notice(message) => match parse_iteration_notice(message) {
+ Some((iteration, phase)) => {
+ self.open_iteration(phase, &format!("iteration {iteration}"))
+ }
+ None => self.note(message),
+ },
Event::ProviderRetrying { attempt, attempts, reason } => {
- self.line("", &format!("retrying ({attempt}/{attempts}): {reason}"));
+ self.note(&format!("retrying ({attempt}/{attempts}): {reason}"));
}
// Only the assistant's. The steer that opens each iteration and the gate's refusals
// arrive as user messages the engine wrote to itself, and the notices above already
@@ -78,7 +122,8 @@ impl LoopView {
for block in &message.content {
match block {
ContentBlock::Text(text) if !text.trim().is_empty() => {
- self.line("· ", &bounded(first_line(text), MAX_SAID));
+ let said = format!("· {}", bounded(first_line(text), MAX_SAID));
+ self.within(&said);
}
ContentBlock::ToolCall { id, name, arguments } => {
let label = luxe_core::describe_call(
@@ -92,49 +137,114 @@ impl LoopView {
}
}
}
- Event::ToolStarted { call_id, name } => {
- if luxe_core::activity::is_bookkeeping(name) {
- return;
- }
- self.tools_this_turn += 1;
- match self.labels.get(call_id) {
- Some(label) if !label.is_empty() => self.line(" ", &format!("{name} {label}")),
- _ => self.line(" ", name),
- }
- }
- Event::ToolEnded { call_id, is_error } if *is_error => {
- let label = self.labels.get(call_id).cloned().unwrap_or_default();
- self.line(" ", format!("failed: {label}").trim_end());
- }
- Event::CompactionDone { tokens_before, tokens_after } => self.line(
- "",
- &format!("compacted {} → {}", thousands(*tokens_before), thousands(*tokens_after)),
- ),
- Event::Error(message) => self.line("", &format!("error: {}", first_line(message))),
- Event::TurnEnded { reason } => {
- let took = self.turn_started.take().map(|at| at.elapsed()).unwrap_or_default();
- let tools = std::mem::take(&mut self.tools_this_turn);
- self.labels.clear();
- let ended = match reason {
- // Every ordinary ending says the same thing; the interesting ones name
- // themselves — a runaway guard, a cancel, a turn that failed.
- luxe_core::StopReason::EndTurn => "turn ended".to_string(),
- other => format!("turn ended: {other:?}"),
- };
- self.line("", &format!("{ended} · {tools} tools · {}", duration(took)));
- }
+ Event::ToolStarted { call_id, name } => self.tool_started(call_id, name),
+ Event::ToolEnded { call_id, is_error } => self.tool_ended(call_id, *is_error),
+ Event::CompactionDone { tokens_before, tokens_after } => self.note(&format!(
+ "compacted {} → {}",
+ thousands(*tokens_before),
+ thousands(*tokens_after)
+ )),
+ Event::Error(message) => self.note(&format!("error: {}", first_line(message))),
+ Event::TurnEnded { reason } => self.turn_ended(reason),
_ => {}
}
}
- /// Note that a turn has begun, so the line closing it can say how long it took.
+ /// Note that a turn has begun, so the line closing it can say how long it took. The run's
+ /// first turn opens its phase here, since no iteration notice comes before it.
pub(crate) fn turn_started(&mut self) {
self.turn_started = Some(Instant::now());
+ if self.quiet {
+ return;
+ }
+ if let Some(phase) = self.start_phase.take() {
+ self.open_iteration(phase, "start");
+ }
}
- /// One line: how long the run has been going, then what happened.
- fn line(&self, indent: &str, message: &str) {
- outln!("{:>9} {indent}{message}", duration(self.started.elapsed()));
+ fn tool_started(&mut self, call_id: &str, name: &str) {
+ if luxe_core::activity::is_bookkeeping(name) {
+ return;
+ }
+ self.tools_this_turn += 1;
+ let label = self.labels.get(call_id).cloned().unwrap_or_default();
+ let depth = self.body_depth();
+ let at = self.started.elapsed();
+ let printed = self.print(depth, &self.paint.tool(name, &label, CallState::Running));
+ let call = RunningCall { printed, at, name: name.to_string(), label, depth };
+ self.running.insert(call_id.to_string(), call);
+ }
+
+ /// Redraws the call's line in its result's color. Where it can't be redrawn, a failure gets a
+ /// line of its own and a success needs none.
+ fn tool_ended(&mut self, call_id: &str, is_error: bool) {
+ let Some(call) = self.running.remove(call_id) else { return };
+ let state = if is_error { CallState::Failed } else { CallState::Succeeded };
+ let line = self.paint.tool(&call.name, &call.label, state);
+ let stamped = self.stamped(call.at, call.depth, &line);
+ let redrawn = self.paint.in_color() && self.screen.rewrite(call.printed, &stamped);
+ if !redrawn && is_error {
+ self.print(self.body_depth(), &line);
+ }
+ }
+
+ fn turn_ended(&mut self, reason: &luxe_core::StopReason) {
+ let took = self.turn_started.take().map(|at| at.elapsed()).unwrap_or_default();
+ let tools = std::mem::take(&mut self.tools_this_turn);
+ self.labels.clear();
+ self.running.clear();
+ let ended = match reason {
+ // Every ordinary ending says the same thing; the interesting ones name
+ // themselves — a runaway guard, a cancel, a turn that failed.
+ luxe_core::StopReason::EndTurn => "turn ended".to_string(),
+ other => format!("turn ended: {other:?}"),
+ };
+ let summary = self.paint.dim(&format!("{ended} · {tools} tools · {}", duration(took)));
+ let depth = match (self.phase, self.in_iteration) {
+ (Some(phase), true) => Depth::Iteration(phase, Edge::Close),
+ _ => self.body_depth(),
+ };
+ self.in_iteration = false;
+ self.print(depth, &summary);
+ }
+
+ /// Opens an iteration, under a heading first when it starts a new phase.
+ fn open_iteration(&mut self, phase: u32, label: &str) {
+ if self.phase != Some(phase) {
+ self.print(Depth::Phase(phase), &self.paint.phase_heading(phase));
+ self.phase = Some(phase);
+ }
+ self.in_iteration = true;
+ self.print(Depth::Iteration(phase, Edge::Open), label);
+ }
+
+ /// A line inside whatever is open.
+ fn within(&mut self, text: &str) {
+ self.print(self.body_depth(), text);
+ }
+
+ /// A hook's or the engine's notice: inside the iteration when one is open.
+ fn note(&mut self, text: &str) {
+ self.within(text);
+ }
+
+ /// The depth a body line sits at now.
+ fn body_depth(&self) -> Depth {
+ match (self.phase, self.in_iteration) {
+ (Some(phase), true) => Depth::Iteration(phase, Edge::Within),
+ (Some(phase), false) => Depth::Phase(phase),
+ (None, _) => Depth::Run,
+ }
+ }
+
+ /// One line: how long the run has been going, the guides, then what happened.
+ fn print(&mut self, depth: Depth, text: &str) -> Printed {
+ let line = self.stamped(self.started.elapsed(), depth, text);
+ self.screen.print(&line)
+ }
+
+ fn stamped(&self, elapsed: std::time::Duration, depth: Depth, text: &str) -> String {
+ format!("{:>9} {}{text}", duration(elapsed), self.paint.guides(depth))
}
}
@@ -218,7 +328,7 @@ mod tests {
/// still built labels would be a memory cost for output nobody asked for.
#[test]
fn a_quiet_view_does_not_even_look() {
- let mut view = LoopView::new(true);
+ let mut view = recording(true);
view.observe(&Event::MessageCommitted {
session_id: "s".into(),
entry_id: luxe_core::session::EntryId::new(),
@@ -232,7 +342,7 @@ mod tests {
/// A turn's tool count belongs to that turn, and starts again with the next one.
#[test]
fn each_turn_counts_its_own_tools() {
- let mut view = LoopView::new(false);
+ let mut view = recording(false);
view.turn_started();
view.observe(&Event::ToolStarted { call_id: "a".into(), name: "read".into() });
view.observe(&Event::ToolStarted { call_id: "b".into(), name: "edit".into() });
@@ -243,14 +353,83 @@ mod tests {
assert!(view.labels.is_empty(), "and does not inherit the last turn's labels");
}
- /// Luxe's own bookkeeping is not work anyone asked to watch.
+ /// luxe's own bookkeeping is not work anyone asked to watch.
#[test]
fn activating_a_tool_is_not_a_step_worth_a_line() {
- let mut view = LoopView::new(false);
+ let mut view = recording(false);
view.observe(&Event::ToolStarted {
call_id: "c".into(),
name: luxe_core::tool::ACTIVATE_TOOLS_NAME.into(),
});
assert_eq!(view.tools_this_turn, 0);
}
+
+ /// A plain view of a run starting on phase 0, its lines kept.
+ fn recording(quiet: bool) -> LoopView {
+ LoopView::with(quiet, Some(0), Paint::plain(), Screen::recording())
+ }
+
+ /// What the view printed, past the time column.
+ fn printed(view: &LoopView) -> Vec {
+ view.screen.lines().iter().map(|line| line[11..].to_string()).collect()
+ }
+
+ fn notice(text: &str) -> Event {
+ Event::Notice(text.into())
+ }
+
+ fn ended(view: &mut LoopView, call_id: &str, is_error: bool) {
+ view.observe(&Event::ToolEnded { call_id: call_id.into(), is_error });
+ }
+
+ /// The first turn opens its phase and iteration itself; the turn's end closes the iteration.
+ #[test]
+ fn the_first_turn_is_bracketed_under_its_phase() {
+ let mut view = recording(false);
+ view.turn_started();
+ view.observe(&Event::ToolStarted { call_id: "a".into(), name: "read".into() });
+ view.observe(&Event::TurnEnded { reason: luxe_core::StopReason::EndTurn });
+ let lines = printed(&view);
+ assert_eq!(lines[..3], ["┃ phase 0", "┃ ┌ start", "┃ │ ○ read"], "{lines:?}");
+ assert!(lines[3].starts_with("┃ └ turn ended · 1 tools"), "{lines:?}");
+ }
+
+ /// An iteration in the same phase opens without a heading; a new phase gets one first.
+ #[test]
+ fn a_new_phase_gets_a_heading_and_the_same_phase_does_not() {
+ let mut view = recording(false);
+ view.turn_started();
+ view.observe(¬ice("iteration 2 · phase 0"));
+ view.observe(¬ice("iteration 3 · phase 1"));
+ let lines = printed(&view);
+ assert_eq!(lines[2..], ["┃ ┌ iteration 2", "┃ phase 1", "┃ ┌ iteration 3"], "{lines:?}");
+ }
+
+ /// Other notices sit inside the open iteration, and between iterations on the phase rail.
+ #[test]
+ fn a_notice_sits_at_the_depth_it_arrives() {
+ let mut view = recording(false);
+ view.turn_started();
+ view.observe(¬ice("phase 0 claimed complete — running the gate"));
+ view.observe(&Event::TurnEnded { reason: luxe_core::StopReason::EndTurn });
+ view.observe(¬ice("phase 0 accepted — now on phase 1"));
+ let lines = printed(&view);
+ assert_eq!(lines[2], "┃ │ phase 0 claimed complete — running the gate", "{lines:?}");
+ assert_eq!(lines[4], "┃ phase 0 accepted — now on phase 1", "{lines:?}");
+ }
+
+ /// Where the line can't be redrawn, a failure still gets a line saying so, and a success
+ /// adds nothing to the call's own line.
+ #[test]
+ fn without_a_redraw_only_a_failure_adds_a_line() {
+ let mut view = recording(false);
+ view.turn_started();
+ view.observe(&Event::ToolStarted { call_id: "a".into(), name: "read".into() });
+ view.observe(&Event::ToolStarted { call_id: "b".into(), name: "bash".into() });
+ ended(&mut view, "a", false);
+ ended(&mut view, "b", true);
+ let lines = printed(&view);
+ assert_eq!(lines.len(), 5, "{lines:?}");
+ assert_eq!(lines[4], "┃ │ ✗ bash · failed", "{lines:?}");
+ }
}
diff --git a/crates/luxe-cli/src/loop_view/paint.rs b/crates/luxe-cli/src/loop_view/paint.rs
new file mode 100644
index 0000000..56364ee
--- /dev/null
+++ b/crates/luxe-cli/src/loop_view/paint.rs
@@ -0,0 +1,193 @@
+//! Paint — how a loop line looks: the guides that lead it, and its colors when there are any.
+//!
+//! The guides work like an editor's indent guides. A phase draws a heavy rail down the left, in
+//! a color of its own; an iteration draws a light one inside it, opened with `┌` and closed with
+//! `└`. Scrolled back through a long run, or seen in a minimap, each phase is a band of one color
+//! and each iteration a bracket within it. Colors come from the terminal's own sixteen, so its
+//! theme picks the shades, and the guides are plain characters, so a log without color keeps them.
+
+use crossterm::style::{Color, Stylize};
+
+/// Phase rail colors, cycled by phase number so two neighboring phases never share one.
+const PHASE_COLORS: [Color; 4] = [Color::Cyan, Color::Magenta, Color::Blue, Color::Yellow];
+
+/// Where a line sits in the run, which decides the guides leading it.
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub(super) enum Depth {
+ /// Before the run has a phase
+ Run,
+ /// Inside a phase, between iterations
+ Phase(u32),
+ /// Inside an iteration of a phase.
+ ///
+ /// Args: the phase, and which edge of the iteration the line is.
+ Iteration(u32, Edge),
+}
+
+/// Which part of an iteration a line is: its opening, its body, or its closing line.
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub(super) enum Edge {
+ /// `┌`, the iteration's first line
+ Open,
+ /// `│`, every line between
+ Within,
+ /// `└`, the turn's summary
+ Close,
+}
+
+/// How a tool call has gone so far.
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub(super) enum CallState {
+ /// Started, no result yet
+ Running,
+ /// Ended without an error
+ Succeeded,
+ /// Ended in an error
+ Failed,
+}
+
+/// Draws loop lines, in color or not.
+#[derive(Clone, Copy, Debug)]
+pub(super) struct Paint {
+ color: bool,
+}
+
+impl Paint {
+ /// Color when stdout is a terminal that wants it: not `NO_COLOR`, not `TERM=dumb`.
+ pub(super) fn detect() -> Self {
+ use std::io::IsTerminal;
+ let no_color = std::env::var_os("NO_COLOR").is_some_and(|value| !value.is_empty());
+ let dumb = std::env::var("TERM").is_ok_and(|term| term == "dumb");
+ Self { color: std::io::stdout().is_terminal() && !no_color && !dumb }
+ }
+
+ /// No color at all, as a pipe or a log gets.
+ #[cfg(test)]
+ pub(super) fn plain() -> Self {
+ Self { color: false }
+ }
+
+ /// Whether this paints in color.
+ pub(super) fn in_color(self) -> bool {
+ self.color
+ }
+
+ /// The guides for a line at `depth`, each followed by a space.
+ pub(super) fn guides(self, depth: Depth) -> String {
+ match depth {
+ Depth::Run => String::new(),
+ Depth::Phase(phase) => format!("{} ", self.rail(phase)),
+ Depth::Iteration(phase, edge) => {
+ format!("{} {} ", self.rail(phase), self.dim(edge_glyph(edge)))
+ }
+ }
+ }
+
+ /// The line that opens a phase: its number in the phase's color.
+ pub(super) fn phase_heading(self, phase: u32) -> String {
+ let text = format!("phase {phase}");
+ match self.color {
+ true => text.with(phase_color(phase)).bold().to_string(),
+ false => text,
+ }
+ }
+
+ /// A tool call: a mark and the tool's name colored by how it went, then what it was asked.
+ /// A failure says so in words too, so the color never carries it alone.
+ pub(super) fn tool(self, name: &str, label: &str, state: CallState) -> String {
+ let (mark, color) = match state {
+ CallState::Running => ("○", None),
+ CallState::Succeeded => ("✓", Some(Color::Green)),
+ CallState::Failed => ("✗", Some(Color::Red)),
+ };
+ let head = format!("{mark} {name}");
+ let head = match (self.color, color) {
+ (true, Some(color)) => head.with(color).bold().to_string(),
+ (true, None) => head.bold().to_string(),
+ (false, _) => head,
+ };
+ let failed = match state {
+ CallState::Failed => format!(" · {}", self.colored("failed", Color::Red)),
+ _ => String::new(),
+ };
+ format!("{head} {label}").trim_end().to_string() + &failed
+ }
+
+ /// Secondary text: a turn's summary, the iteration rail.
+ pub(super) fn dim(self, text: &str) -> String {
+ match self.color {
+ true => text.dim().to_string(),
+ false => text.to_string(),
+ }
+ }
+
+ fn colored(self, text: &str, color: Color) -> String {
+ match self.color {
+ true => text.with(color).to_string(),
+ false => text.to_string(),
+ }
+ }
+
+ fn rail(self, phase: u32) -> String {
+ self.colored("┃", phase_color(phase))
+ }
+}
+
+fn phase_color(phase: u32) -> Color {
+ PHASE_COLORS[phase as usize % PHASE_COLORS.len()]
+}
+
+fn edge_glyph(edge: Edge) -> &'static str {
+ match edge {
+ Edge::Open => "┌",
+ Edge::Within => "│",
+ Edge::Close => "└",
+ }
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ #[test]
+ fn each_depth_draws_its_own_guides() {
+ let paint = Paint::plain();
+ assert_eq!(paint.guides(Depth::Run), "");
+ assert_eq!(paint.guides(Depth::Phase(2)), "┃ ");
+ assert_eq!(paint.guides(Depth::Iteration(2, Edge::Open)), "┃ ┌ ");
+ assert_eq!(paint.guides(Depth::Iteration(2, Edge::Within)), "┃ │ ");
+ assert_eq!(paint.guides(Depth::Iteration(2, Edge::Close)), "┃ └ ");
+ }
+
+ /// Without color a failure still reads as one, and a success has no word to need.
+ #[test]
+ fn a_failed_call_says_so_in_words() {
+ let paint = Paint::plain();
+ assert_eq!(paint.tool("read", "a.rs", CallState::Running), "○ read a.rs");
+ assert_eq!(paint.tool("read", "a.rs", CallState::Succeeded), "✓ read a.rs");
+ assert_eq!(
+ paint.tool("bash", "cargo test", CallState::Failed),
+ "✗ bash cargo test · failed"
+ );
+ assert_eq!(paint.tool("todo", "", CallState::Succeeded), "✓ todo", "no trailing space");
+ }
+
+ /// The tool's name takes the result's color; a plain view has no escapes at all.
+ #[test]
+ fn the_name_is_colored_by_the_result() {
+ let paint = Paint { color: true };
+ let ok = paint.tool("read", "a.rs", CallState::Succeeded);
+ let failed = paint.tool("read", "a.rs", CallState::Failed);
+ assert!(ok.contains(&"✓ read".with(Color::Green).bold().to_string()), "{ok:?}");
+ assert!(failed.contains(&"✗ read".with(Color::Red).bold().to_string()), "{failed:?}");
+ assert!(!Paint::plain().tool("read", "a.rs", CallState::Failed).contains('\x1b'));
+ }
+
+ /// Neighboring phases never share a rail color, so a phase boundary shows in a minimap.
+ #[test]
+ fn neighboring_phases_differ() {
+ for phase in 0..8 {
+ assert_ne!(phase_color(phase), phase_color(phase + 1), "{phase}");
+ }
+ }
+}
diff --git a/crates/luxe-cli/src/loop_view/screen.rs b/crates/luxe-cli/src/loop_view/screen.rs
new file mode 100644
index 0000000..3b1f915
--- /dev/null
+++ b/crates/luxe-cli/src/loop_view/screen.rs
@@ -0,0 +1,177 @@
+//! Screen — where loop lines go, and taking one back to redraw it.
+//!
+//! A tool call is printed when it starts, because "is it still going?" is the first question about
+//! an unattended run, and its result only exists once it ends. On a terminal the start line is
+//! redrawn in place with the result. That's only safe when every line is exactly one row, so on a
+//! terminal each line is cut to the width rather than left to wrap, and a resize since the line
+//! was printed means it's left alone. A pipe or a file never gets a redraw.
+
+use crate::output::{out, outln};
+use unicode_width::UnicodeWidthChar;
+
+/// A line already printed, and what's needed to find it again.
+#[derive(Clone, Copy, Debug)]
+pub(super) struct Printed {
+ /// How many lines came before it
+ index: u64,
+ /// The terminal's width when it was printed
+ width: Option,
+}
+
+/// The terminal or pipe the view prints to.
+pub(super) struct Screen {
+ /// Whether lines may be cut to the width and redrawn
+ terminal: bool,
+ /// Lines printed so far
+ printed: u64,
+ /// Where a test's lines go instead of stdout
+ recorded: Option>,
+}
+
+impl Screen {
+ /// A terminal when stdout is one.
+ pub(super) fn detect() -> Self {
+ use std::io::IsTerminal;
+ Self { terminal: std::io::stdout().is_terminal(), printed: 0, recorded: None }
+ }
+
+ /// A pipe that keeps its lines for a test to read.
+ #[cfg(test)]
+ pub(super) fn recording() -> Self {
+ Self { terminal: false, printed: 0, recorded: Some(Vec::new()) }
+ }
+
+ /// What a [`Screen::recording`] was given.
+ #[cfg(test)]
+ pub(super) fn lines(&self) -> &[String] {
+ self.recorded.as_deref().unwrap_or_default()
+ }
+
+ /// Prints `line` as one row, and says where it went.
+ pub(super) fn print(&mut self, line: &str) -> Printed {
+ let width = self.width();
+ let printed = Printed { index: self.printed, width };
+ self.printed += 1;
+ if let Some(recorded) = &mut self.recorded {
+ recorded.push(line.to_string());
+ return printed;
+ }
+ match width {
+ Some(cells) => outln!("{}", fit(line, usize::from(cells))),
+ None => outln!("{line}"),
+ }
+ printed
+ }
+
+ /// Redraws an earlier line in place. False, and nothing written, when it can't be trusted to
+ /// land on the same row: not a terminal, a resize since, or scrolled past the top.
+ pub(super) fn rewrite(&mut self, at: Printed, line: &str) -> bool {
+ let Some(cells) = self.width().filter(|now| Some(*now) == at.width) else { return false };
+ let up = self.printed - at.index;
+ let rows = crossterm::terminal::size().map_or(0, |(_, rows)| u64::from(rows));
+ if up == 0 || up >= rows {
+ return false;
+ }
+ // Up to the line, clear it, draw, then back down to where the next line goes
+ out!("\x1b[{up}A\r\x1b[2K{}\x1b[{up}B\r", fit(line, usize::from(cells)));
+ true
+ }
+
+ fn width(&self) -> Option {
+ match self.terminal {
+ true => crossterm::terminal::size().ok().map(|(cols, _)| cols),
+ false => None,
+ }
+ }
+}
+
+/// `line` cut to fit in `width` cells, so it never wraps. Escape sequences pass through uncounted,
+/// and a cut line ends in `…` and, if it had color, a reset.
+pub(super) fn fit(line: &str, width: usize) -> String {
+ if visible_width(line) <= width {
+ return line.to_string();
+ }
+ let (cut, _) = scan(line, width.saturating_sub(1));
+ let reset = if line.contains('\x1b') { "\x1b[0m" } else { "" };
+ format!("{cut}…{reset}")
+}
+
+/// Where the scanner is: in text, just past an escape, or inside a `CSI` sequence.
+#[derive(Clone, Copy)]
+enum Scan {
+ Text,
+ Escape,
+ Csi,
+}
+
+/// `line` up to `budget` visible cells, with every escape sequence kept, and the cells it used.
+fn scan(line: &str, budget: usize) -> (String, usize) {
+ let mut kept = String::with_capacity(line.len());
+ let mut used = 0;
+ let mut state = Scan::Text;
+ for ch in line.chars() {
+ state = match (state, ch) {
+ (Scan::Text, '\x1b') => Scan::Escape,
+ (Scan::Escape, '[') => Scan::Csi,
+ (Scan::Escape, _) => Scan::Text,
+ (Scan::Csi, '@'..='~') => Scan::Text,
+ (Scan::Csi, _) => Scan::Csi,
+ (Scan::Text, text) => {
+ let cells = text.width().unwrap_or(0);
+ if used + cells > budget {
+ break;
+ }
+ used += cells;
+ Scan::Text
+ }
+ };
+ kept.push(ch);
+ }
+ (kept, used)
+}
+
+fn visible_width(line: &str) -> usize {
+ scan(line, usize::MAX).1
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ #[test]
+ fn a_line_that_fits_is_left_alone() {
+ assert_eq!(fit("○ read a.rs", 40), "○ read a.rs");
+ assert_eq!(fit("exactly", 7), "exactly");
+ }
+
+ /// One cell short of the width plus the ellipsis is the width, so the row never wraps.
+ #[test]
+ fn a_long_line_is_cut_to_the_width() {
+ assert_eq!(fit("abcdefghij", 6), "abcde…");
+ assert_eq!(visible_width(&fit("abcdefghij", 6)), 6);
+ }
+
+ /// Escapes take no cells: a colored line that fits isn't cut, and a cut one is reset.
+ #[test]
+ fn escapes_are_not_counted() {
+ let green = "\x1b[32m✓ read\x1b[39m a.rs";
+ assert_eq!(visible_width(green), "✓ read a.rs".chars().count());
+ assert_eq!(fit(green, 11), green);
+ assert_eq!(fit(green, 5), "\x1b[32m✓ re…\x1b[0m");
+ }
+
+ /// A wide character counts as the two cells it takes.
+ #[test]
+ fn wide_characters_count_double() {
+ assert_eq!(visible_width("日本"), 4);
+ assert_eq!(fit("日本語", 5), "日本…");
+ }
+
+ /// A pipe never redraws: there's no row to go back to.
+ #[test]
+ fn a_pipe_never_redraws() {
+ let mut screen = Screen::recording();
+ let line = screen.print("○ read a.rs");
+ assert!(!screen.rewrite(line, "✓ read a.rs"));
+ }
+}
diff --git a/crates/luxe-cli/src/model_catalog.rs b/crates/luxe-cli/src/model_catalog.rs
index df70c35..2647679 100644
--- a/crates/luxe-cli/src/model_catalog.rs
+++ b/crates/luxe-cli/src/model_catalog.rs
@@ -1,7 +1,7 @@
//! Model catalog wiring — cache location, refresh scheduling, and the `/model` menu.
//!
//! Metadata itself is resolved by [`luxe_providers::catalog`]; this module owns the parts that
-//! depend on where Luxe keeps files and when it's allowed to touch the network. Resolution is
+//! depend on where luxe keeps files and when it's allowed to touch the network. Resolution is
//! always offline-safe — it reads the embedded snapshot plus a local cache and never blocks —
//! while [`spawn_refresh`] keeps that cache current in the background.
@@ -71,7 +71,7 @@ pub(crate) fn build_menu(settings: &AgentSettings) -> ModelMenu {
push(&mut entries, resolve_model(&config, provider, id).0, true);
}
- // Then everything the catalog knows for providers Luxe can actually route to: the ones with a
+ // Then everything the catalog knows for providers luxe can actually route to: the ones with a
// config block, anything the user has signed in to (a Copilot login is worthless if its models
// never appear), and the launch's own provider when the launch itself carries the credentials
// to reach it.
diff --git a/crates/luxe-cli/src/model_catalog/tests.rs b/crates/luxe-cli/src/model_catalog/tests.rs
index 5143134..08f3a24 100644
--- a/crates/luxe-cli/src/model_catalog/tests.rs
+++ b/crates/luxe-cli/src/model_catalog/tests.rs
@@ -187,7 +187,7 @@ fn an_unconfigured_machine_has_no_models_to_choose_between() {
let with_key = AgentSettings::resolve(std::path::PathBuf::from("."), Default::default());
let catalog = build_menu(&with_key).catalog;
std::env::remove_var("ANTHROPIC_API_KEY");
- assert!(!catalog.is_empty(), "an ambient key is a provider Luxe can route to");
+ assert!(!catalog.is_empty(), "an ambient key is a provider luxe can route to");
assert!(
catalog.iter().all(|entry| entry.info.r#ref.provider == "anthropic"),
"and only that provider's models: {:?}",
diff --git a/crates/luxe-cli/src/output.rs b/crates/luxe-cli/src/output.rs
index 83c9c13..f6fd11b 100644
--- a/crates/luxe-cli/src/output.rs
+++ b/crates/luxe-cli/src/output.rs
@@ -5,9 +5,9 @@
//! idiom.
//!
//! The usual cure — resetting `SIGPIPE` to its default so the process dies silently — is wrong
-//! here. Luxe writes to hook, MCP and LSP pipes and relies on `EPIPE` arriving as an error: the
+//! here. luxe writes to hook, MCP and LSP pipes and relies on `EPIPE` arriving as an error: the
//! command-hook runner deliberately tolerates a hook that exits without reading its stdin, and a
-//! default `SIGPIPE` would kill Luxe at that line instead. So the pipe stays ignored and the
+//! default `SIGPIPE` would kill luxe at that line instead. So the pipe stays ignored and the
//! writing goes through one place, where a reader that has gone away is a quiet, successful end of
//! output.
//!
diff --git a/crates/luxe-cli/src/pull_request_watch.rs b/crates/luxe-cli/src/pull_request_watch.rs
index c65edf8..bcae29c 100644
--- a/crates/luxe-cli/src/pull_request_watch.rs
+++ b/crates/luxe-cli/src/pull_request_watch.rs
@@ -249,10 +249,10 @@ async fn look(
fn unavailable(error: &GitHubError) -> Option {
let reason = match error {
GitHubError::NotInstalled => {
- "The GitHub CLI (`gh`) is not installed, so Luxe cannot see this branch's pull request."
+ "The GitHub CLI (`gh`) is not installed, so luxe cannot see this branch's pull request."
}
GitHubError::NotAuthenticated { .. } => {
- "The GitHub CLI is not signed in. Run `gh auth login`, and Luxe picks it up the next \
+ "The GitHub CLI is not signed in. Run `gh auth login`, and luxe picks it up the next \
time the terminal has focus."
}
GitHubError::NotGitHub => "This repository has no remote on GitHub.",
diff --git a/crates/luxe-cli/src/review_agent.rs b/crates/luxe-cli/src/review_agent.rs
index 208f9e6..46360b3 100644
--- a/crates/luxe-cli/src/review_agent.rs
+++ b/crates/luxe-cli/src/review_agent.rs
@@ -1,6 +1,6 @@
//! The reviewer as an agent — a real turn, with the real tools, in a session of its own.
//!
-//! A review is a model driving Luxe. It gets the same pipeline every other turn gets: the tool
+//! A review is a model driving luxe. It gets the same pipeline every other turn gets: the tool
//! registry, the permission engine, per-tool deadlines, cancellation, the hook chain. What makes it
//! a *review* is how its engine is built — the reviewer's model, the reviewer's tools and no others,
//! and a mode that refuses every write — and the brief it is handed, which says who it is.
diff --git a/crates/luxe-cli/src/workspace_data.rs b/crates/luxe-cli/src/workspace_data.rs
index 9a67be8..9291373 100644
--- a/crates/luxe-cli/src/workspace_data.rs
+++ b/crates/luxe-cli/src/workspace_data.rs
@@ -313,7 +313,7 @@ fn bounded_text(text: &str, limit: usize) -> String {
value
}
-/// Luxe's own project directory, excluded from what it reports as *your* changes.
+/// luxe's own project directory, excluded from what it reports as *your* changes.
const LUXE_PROJECT_DIR: &str = ".luxe";
async fn project_identity(root: &Path) -> ProjectIdentity {
diff --git a/crates/luxe-cli/src/workspace_data/changes.rs b/crates/luxe-cli/src/workspace_data/changes.rs
index ef30555..9020101 100644
--- a/crates/luxe-cli/src/workspace_data/changes.rs
+++ b/crates/luxe-cli/src/workspace_data/changes.rs
@@ -17,7 +17,7 @@ pub(super) async fn collect_change_summary(root: PathBuf) -> WorkspaceChangesSum
let counts = numstat.as_deref().map(parse_numstat).unwrap_or_default();
let files = parse_status(&status)
.into_iter()
- // Luxe's own project directory — sessions, logs, tasks, review artifacts — is not the
+ // luxe's own project directory — sessions, logs, tasks, review artifacts — is not the
// user's work. Listing it made the dock's count disagree with the review's forever.
.filter(|(path, _)| !path.starts_with(LUXE_PROJECT_DIR))
.map(|(path, status)| {
diff --git a/crates/luxe-cli/src/workspace_data/external.rs b/crates/luxe-cli/src/workspace_data/external.rs
index 178267f..a44efbc 100644
--- a/crates/luxe-cli/src/workspace_data/external.rs
+++ b/crates/luxe-cli/src/workspace_data/external.rs
@@ -32,7 +32,7 @@ pub(super) async fn collect_external_summary(
}
let changes = collect_change_summary(root.clone()).await;
// A repository that *contains* the project — an edit in a sibling directory of a checkout
- // Luxe was opened one level inside — reports the project's own files too. Those already
+ // luxe was opened one level inside — reports the project's own files too. Those already
// have a section, and listing them twice would double every number on the screen.
let mut files = changes.files;
files.retain(|file| !luxe_core::paths::contains(project_root, &root.join(&file.path)));
diff --git a/crates/luxe-cli/src/workspace_data/tests.rs b/crates/luxe-cli/src/workspace_data/tests.rs
index fe308a9..a810412 100644
--- a/crates/luxe-cli/src/workspace_data/tests.rs
+++ b/crates/luxe-cli/src/workspace_data/tests.rs
@@ -118,7 +118,7 @@ fn git_repo_with_a_commit(root: &Path) {
}
/// A checkout is asked the same question the project is asked, which is the whole point of
-/// recording the root instead of the file: this change is made with the shell, so nothing Luxe
+/// recording the root instead of the file: this change is made with the shell, so nothing luxe
/// tracks per-file could have seen it.
#[tokio::test]
async fn an_external_checkout_is_reported_by_git() {
@@ -130,7 +130,7 @@ async fn an_external_checkout_is_reported_by_git() {
let work = ExternalWork::new();
work.observe(&sibling.join("lib.rs"));
- // Not through any Luxe tool — exactly what the per-file ledger cannot know about.
+ // Not through any luxe tool — exactly what the per-file ledger cannot know about.
std::fs::write(sibling.join("lib.rs"), "alpha\nbeta\ngamma\n").unwrap();
let summary = collect_external_summary(work.record(), &project).await;
@@ -143,7 +143,7 @@ async fn an_external_checkout_is_reported_by_git() {
assert_eq!(summary.total_files, 1);
}
-/// A repository Luxe was opened *inside* reports the project's own files too. They already have
+/// A repository luxe was opened *inside* reports the project's own files too. They already have
/// a section, and counting them twice would double every number on screen.
#[tokio::test]
async fn a_parent_checkout_does_not_report_the_project_twice() {
diff --git a/crates/luxe-cli/src/workspace_runtime/tests.rs b/crates/luxe-cli/src/workspace_runtime/tests.rs
index f84e409..3610c3c 100644
--- a/crates/luxe-cli/src/workspace_runtime/tests.rs
+++ b/crates/luxe-cli/src/workspace_runtime/tests.rs
@@ -119,7 +119,7 @@ async fn bridge_tracks_real_engine_startup_and_new_sessions() {
external_work: None,
review_context_files: Vec::new(),
review_exclude: Vec::new(),
- remember_runtime_state: false,
+ remember_choices_in: None,
cache_warming: luxe_core::cache_warming::CacheWarming::Off,
review_insight_provider: None,
github: None,
diff --git a/crates/luxe-cli/tests/mcp_commands/main.rs b/crates/luxe-cli/tests/mcp_commands/main.rs
index d39710a..2301be1 100644
--- a/crates/luxe-cli/tests/mcp_commands/main.rs
+++ b/crates/luxe-cli/tests/mcp_commands/main.rs
@@ -15,15 +15,15 @@ fn project_server(project: &Path) {
std::fs::write(project.join(".mcp.json"), config.to_string()).unwrap();
}
-/// Whether Luxe launched the project's server. The marker is the proof when the child gets that far,
-/// but the child is this test binary starting up again, and on a slow runner Luxe can give up on the
+/// Whether luxe launched the project's server. The marker is the proof when the child gets that far,
+/// but the child is this test binary starting up again, and on a slow runner luxe can give up on the
/// closed connection before that startup reaches the write; its report of the attempt is the proof
/// then. A refused server is never launched, so neither can appear.
fn started(project: &Path, stdout: &str) -> bool {
project.join("mcp-ran").exists() || stdout.contains("project-marker: failed to connect")
}
-/// Whether Luxe refused to start the project's server: it says so, naming it. Checked alongside the
+/// Whether luxe refused to start the project's server: it says so, naming it. Checked alongside the
/// absent marker, which alone would also pass for a server launched too slowly to have written it.
fn refused(project: &Path, stdout: &str) -> bool {
stdout.contains("Not starting project MCP servers: project-marker")
diff --git a/crates/luxe-cli/tests/rpc_stdio.rs b/crates/luxe-cli/tests/rpc_stdio.rs
index 059b9d8..b10995d 100644
--- a/crates/luxe-cli/tests/rpc_stdio.rs
+++ b/crates/luxe-cli/tests/rpc_stdio.rs
@@ -272,7 +272,7 @@ fn changed_repository() -> tempfile::TempDir {
};
git(&["init", "--quiet", "--initial-branch=main"]);
git(&["config", "user.email", "tests@example.invalid"]);
- git(&["config", "user.name", "Luxe Tests"]);
+ git(&["config", "user.name", "luxe Tests"]);
std::fs::write(project.path().join("total.rs"), "fn total() -> u32 {\n 1\n}\n").unwrap();
git(&["add", "total.rs"]);
git(&["commit", "--quiet", "-m", "initial"]);
diff --git a/crates/luxe-core/Cargo.toml b/crates/luxe-core/Cargo.toml
index dc1e847..8acb24f 100644
--- a/crates/luxe-core/Cargo.toml
+++ b/crates/luxe-core/Cargo.toml
@@ -3,7 +3,7 @@ name = "luxe-core"
version.workspace = true
edition.workspace = true
license.workspace = true
-description = "Luxe's engine: the agent loop, session model, context assembly, permissions, and the Provider/Tool/Hook traits"
+description = "luxe's engine: the agent loop, session model, context assembly, permissions, and the Provider/Tool/Hook traits"
repository.workspace = true
homepage.workspace = true
@@ -66,7 +66,7 @@ globset = "0.4"
[target.'cfg(unix)'.dependencies]
rustix = { version = "1", features = ["process", "stdio", "system"] }
-# Pointing stdin at the null device on Windows, and keeping the pipe Luxe took out of every child:
+# Pointing stdin at the null device on Windows, and keeping the pipe luxe took out of every child:
# no safe API does either, and without it every child of `luxe --rpc` stalls behind the protocol
# read. See `program::take_stdin`. Job objects end a command's whole tree on a timeout, where
# Windows has no process group to signal; see `program::TreeGuard`. A session's named pipe needs a
diff --git a/crates/luxe-core/src/activity.rs b/crates/luxe-core/src/activity.rs
index ba58213..9983449 100644
--- a/crates/luxe-core/src/activity.rs
+++ b/crates/luxe-core/src/activity.rs
@@ -211,7 +211,7 @@ impl ToolActivity {
}
}
-/// Whether a tool is Luxe's own housekeeping rather than work someone asked for.
+/// Whether a tool is luxe's own housekeeping rather than work someone asked for.
///
/// Only [`crate::tool::ACTIVATE_TOOLS_NAME`], and only because *we* tier the registry: the model
/// asks for a deferred tool's schema before it can call it, which is a fact about our request
diff --git a/crates/luxe-core/src/auth.rs b/crates/luxe-core/src/auth.rs
index 33e9e7f..8808403 100644
--- a/crates/luxe-core/src/auth.rs
+++ b/crates/luxe-core/src/auth.rs
@@ -189,7 +189,7 @@ pub enum LoginScope {
/// Every configurable provider, with every method it offers.
#[default]
Any,
- /// Providers whose subscription Luxe is allowed to sign in to.
+ /// Providers whose subscription luxe is allowed to sign in to.
Subscription,
/// Providers that take a key you already have.
ApiKey,
@@ -360,8 +360,6 @@ pub struct LoginOutcome {
/// A local server takes no credential, and "API key" would be a claim the file contradicts.
pub kind: Option,
pub account: Option,
- /// A model id worth defaulting to, when the flow learned one.
- pub suggested_model: Option,
/// Factual summary lines about what now exists (e.g. "17 models available").
pub notes: Vec,
/// Advice that only makes sense once the credential exists, shown apart from the summary.
diff --git a/crates/luxe-core/src/briefs.rs b/crates/luxe-core/src/briefs.rs
index f21549c..1c70f14 100644
--- a/crates/luxe-core/src/briefs.rs
+++ b/crates/luxe-core/src/briefs.rs
@@ -1,7 +1,7 @@
-//! Briefs — the messages Luxe writes to the agent itself, told apart from the reader's own.
+//! Briefs — the messages luxe writes to the agent itself, told apart from the reader's own.
//!
//! A draft, a question from a review and a batch of review comments each reach the agent as a
-//! user message Luxe composed: the instructions, the evidence, the IDs a report has to echo. The
+//! user message luxe composed: the instructions, the evidence, the IDs a report has to echo. The
//! transcript is the reader's conversation, and a page of plumbing in their own colors reads as
//! something they said. So a front-end shows one line saying what happened instead — or, for a
//! question from a review, just the question, since those words *are* the reader's — and the
@@ -90,7 +90,7 @@ const FINDING_ID: &str = " Finding ID: ";
const EVIDENCE: &str = " Evidence: ";
const REVIEWER: &str = " Reviewer: ";
/// The lines that follow a comment's body; none of them is the reader's question.
-const AFTER_BODY: [&str; 4] = [" Source: ", " Proposed fix: ", " Agent: ", " Luxe: "];
+const AFTER_BODY: [&str; 4] = [" Source: ", " Proposed fix: ", " Agent: ", " luxe: "];
/// How review feedback closes.
const CLOSING: &str = "Do not rewrite unrelated code.";
@@ -158,7 +158,7 @@ pub fn review_questions(text: &str) -> Vec {
/// How a draft's fence opens, its newline included.
const DRAFT_FENCE: &str = "```luxe-draft\n";
-/// An answer as the reader sees it: without what the agent wrote for Luxe to read.
+/// An answer as the reader sees it: without what the agent wrote for luxe to read.
///
/// A fix's remediation report goes, since the review shows it comment by comment, and the prose
/// around it stays. An answer carrying a draft goes whole: the draft opens in the card or composer
diff --git a/crates/luxe-core/src/cache_warming.rs b/crates/luxe-core/src/cache_warming.rs
index f5dd60f..f4f6111 100644
--- a/crates/luxe-core/src/cache_warming.rs
+++ b/crates/luxe-core/src/cache_warming.rs
@@ -14,7 +14,7 @@
//!
//! # Design
//!
-//! Same policy as Pi's, checked against Luxe's own ledgers:
+//! Same policy as Pi's, checked against luxe's own ledgers:
//!
//! - **While a turn runs** the next request is certain, so a refresh pays whenever a read costs
//! less than the write it saves. Across 2,980 recorded Claude calls this rescued 12 of 14 waits
diff --git a/crates/luxe-core/src/compaction/budget.rs b/crates/luxe-core/src/compaction/budget.rs
index e5712b9..9610117 100644
--- a/crates/luxe-core/src/compaction/budget.rs
+++ b/crates/luxe-core/src/compaction/budget.rs
@@ -32,7 +32,7 @@ pub fn estimate_message_tokens(messages: &[Message]) -> u64 {
/// Estimate the tokens a request spends before any conversation: the system prompt and the tool
/// schemas.
///
-/// Far from negligible — Luxe's harness plus two dozen JSON schemas is tens of thousands of
+/// Far from negligible — luxe's harness plus two dozen JSON schemas is tens of thousands of
/// tokens, and leaving it out of a prompt figure is how a request that "fits" turns into a 400.
pub fn estimate_prefix_tokens(system: Option<&str>, tools: &[ToolSchema]) -> u64 {
let system = system.map(|text| text.len() as u64 / 4).unwrap_or(0);
@@ -176,7 +176,7 @@ pub fn usable_context_window(context_window: u64, max_tokens: u32, reserve_token
/// percent on a 200K prompt is thousands of tokens out. Mirrors Pi's margin.
pub const CONTEXT_SAFETY_TOKENS: u32 = 4_096;
-/// The smallest output cap Luxe will ask for once the window is full. Useless as an answer, but a
+/// The smallest output cap luxe will ask for once the window is full. Useless as an answer, but a
/// request that asks for zero (or a negative) output is a hard rejection rather than a short reply.
const MIN_MAX_TOKENS: u32 = 1;
diff --git a/crates/luxe-core/src/config.rs b/crates/luxe-core/src/config.rs
index 192239b..66cf044 100644
--- a/crates/luxe-core/src/config.rs
+++ b/crates/luxe-core/src/config.rs
@@ -2,7 +2,7 @@
//!
//! Loads `~/.config/luxe/config.json` (global) and `/.luxe/config.json`
//! (project-local), merges them (project wins), and falls back to defaults on any
-//! parse error. Unknown fields are ignored so a newer Luxe's config keeps working
+//! parse error. Unknown fields are ignored so a newer luxe's config keeps working
//! with an older build and vice versa.
//!
//! The schema is published via [`Config::schema_json`] for editor autocomplete
@@ -31,7 +31,7 @@ pub use integrations::*;
use merge::{merge_color_overrides, merge_disabled};
pub use write::{value_at, ConfigLayers, ConfigWriteError, Layer};
-/// The theme a launch lands on when nothing else decides: Luxe's own.
+/// The theme a launch lands on when nothing else decides: luxe's own.
///
/// A default theme is the product's face, so it is the one drawn for it rather than a neutral
/// stand-in. An explicit setting and a matching desktop theme both still win.
@@ -75,7 +75,7 @@ pub struct Config {
/// use default (silent), because a modern terminal turns a bell into a window-manager attention
/// request, which pulls focus off whatever the reader walked away to do.
pub bell: Option,
- /// Report the agent's state to herdr when Luxe runs in one of its panes. `None` = use default
+ /// Report the agent's state to herdr when luxe runs in one of its panes. `None` = use default
/// (on).
pub herdr: Option,
/// Let another front-end on this machine join a running session. `None` = use default (on).
@@ -105,9 +105,9 @@ pub struct Config {
pub compaction: CompactionConfig,
/// Language-server behavior
pub lsp: LspConfig,
- /// Whether Luxe checks for a newer release
+ /// Whether luxe checks for a newer release
pub update: UpdateConfig,
- /// How Luxe keeps up with GitHub
+ /// How luxe keeps up with GitHub
pub github: GithubConfig,
/// Whether sessions work in worktrees of their own
pub worktree: WorktreeConfig,
@@ -137,7 +137,7 @@ impl Config {
}
}
- /// Whether Luxe opens on the start screen (default: yes).
+ /// Whether luxe opens on the start screen (default: yes).
pub fn dashboard_shown(&self) -> bool {
!matches!(self.dashboard, Some(DashboardSetting::Enabled(false)))
}
@@ -186,7 +186,7 @@ impl Config {
self.notifications.as_ref().is_some_and(Notifications::any)
}
- /// Whether the agent's state is reported to herdr when Luxe runs inside it (default: yes).
+ /// Whether the agent's state is reported to herdr when luxe runs inside it (default: yes).
pub fn herdr(&self) -> bool {
self.herdr.unwrap_or(true)
}
@@ -202,7 +202,7 @@ impl Config {
/// Resolves the active theme name, falling back to the current desktop theme when known.
///
- /// An explicit Luxe setting always wins. Without one, the desktop's theme is used when the
+ /// An explicit luxe setting always wins. Without one, the desktop's theme is used when the
/// registry carries one (Omarchy's, see [`crate::omarchy`]), because matching the desktop beats
/// showing off, and anything else lands on [`DEFAULT_THEME`].
pub fn resolved_theme_name(&self, registry: &ThemeRegistry) -> String {
@@ -242,7 +242,7 @@ impl Config {
resolved
}
- /// The JSON Schema for the config files: the checked-in `config.schema.json`, which Luxe writes
+ /// The JSON Schema for the config files: the checked-in `config.schema.json`, which luxe writes
/// beside the user's config so an editor can complete and check every key.
pub fn schema_json() -> &'static str {
include_str!("config/config.schema.json")
@@ -349,7 +349,7 @@ mod activity_bar_config_tests {
assert_eq!(config.theme.as_deref(), Some("nord"), "and the rest of the file still lands");
}
- /// A source's kind is the one set, or its name folded; it may not take one of Luxe's own tabs',
+ /// A source's kind is the one set, or its name folded; it may not take one of luxe's own tabs',
/// and it needs a program to run.
#[test]
fn a_mention_source_has_a_kind_of_its_own() {
diff --git a/crates/luxe-core/src/config/config.schema.json b/crates/luxe-core/src/config/config.schema.json
index c0b20f5..cb9464b 100644
--- a/crates/luxe-core/src/config/config.schema.json
+++ b/crates/luxe-core/src/config/config.schema.json
@@ -1,6 +1,6 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
- "title": "Luxe config",
+ "title": "luxe config",
"type": "object",
"properties": {
"$schema": {
@@ -9,7 +9,7 @@
},
"theme": {
"type": "string",
- "description": "Theme name: a built-in (luxe, catppuccin, nord, tokyo-night, and more — /theme lists them all), the name of a file in themes/, or omarchy to follow the Omarchy desktop's theme. When omitted, Luxe follows Omarchy when it is running there, and uses luxe otherwise."
+ "description": "Theme name: a built-in (luxe, catppuccin, nord, tokyo-night, and more — /theme lists them all), the name of a file in themes/, or omarchy to follow the Omarchy desktop's theme. When omitted, luxe follows Omarchy when it is running there, and uses luxe otherwise."
},
"colors": {
"type": "object",
@@ -54,7 +54,7 @@
"type": "string",
"enum": ["auto", "nerd", "nerd_wide", "unicode", "ascii"],
"default": "auto",
- "description": "How much glyph fidelity the TUI asks for. Icons are decoration \u2014 the words beside them carry the meaning \u2014 so every setting is fully usable. nerd draws Material Design icons from a Nerd Font (the look Luxe ships with); nerd_wide draws the same icons with a second cell of clearance, for a Nerd Font's non-Mono cut, whose icons carry two cells of ink behind a one-cell advance and so paint over the space after them; unicode uses Geometric Shapes and Arrows only, so no patched font is needed; ascii reduces to plain punctuation for a minimal or non-UTF-8 terminal. auto (default) picks nerd where those glyphs can actually be drawn \u2014 kitty bundles its own symbol font, and elsewhere fontconfig is asked whether any installed font provides them \u2014 and unicode otherwise. auto never picks nerd_wide: both cuts of a face report the same advance, so there is nothing to measure."
+ "description": "How much glyph fidelity the TUI asks for. Icons are decoration \u2014 the words beside them carry the meaning \u2014 so every setting is fully usable. nerd draws Material Design icons from a Nerd Font (the look luxe ships with); nerd_wide draws the same icons with a second cell of clearance, for a Nerd Font's non-Mono cut, whose icons carry two cells of ink behind a one-cell advance and so paint over the space after them; unicode uses Geometric Shapes and Arrows only, so no patched font is needed; ascii reduces to plain punctuation for a minimal or non-UTF-8 terminal. auto (default) picks nerd where those glyphs can actually be drawn \u2014 kitty bundles its own symbol font, and elsewhere fontconfig is asked whether any installed font provides them \u2014 and unicode otherwise. auto never picks nerd_wide: both cuts of a face report the same advance, so there is nothing to measure."
},
"images": {
"type": "string",
@@ -108,7 +108,7 @@
"herdr": {
"type": "boolean",
"default": true,
- "description": "When Luxe runs in a herdr pane, report whether it is working, idle or waiting on you, so herdr's sidebar, notifications and `herdr agent wait` follow it. On by default; nothing is sent outside herdr."
+ "description": "When luxe runs in a herdr pane, report whether it is working, idle or waiting on you, so herdr's sidebar, notifications and `herdr agent wait` follow it. On by default; nothing is sent outside herdr."
},
"joinable": {
"type": "boolean",
@@ -147,7 +147,7 @@
"type": "array",
"items": { "type": "string" },
"default": [],
- "description": "Project-relative globs a review leaves out, e.g. [\"PROGRESS.md\", \"docs/design/**\"]. Luxe's own .luxe/ directory is always excluded and needs no entry. Supports *, ** and ?; a trailing / matches a directory and everything under it. Excluded files are counted and named in the Brief, never silently dropped."
+ "description": "Project-relative globs a review leaves out, e.g. [\"PROGRESS.md\", \"docs/design/**\"]. luxe's own .luxe/ directory is always excluded and needs no entry. Supports *, ** and ?; a trailing / matches a directory and everything under it. Excluded files are counted and named in the Brief, never silently dropped."
}
},
"additionalProperties": false
@@ -183,7 +183,7 @@
"remember_model": {
"type": "boolean",
"default": true,
- "description": "Remember the model and thinking level picked in a session and use them next launch. Precedence: --model/--thinking flag, then LUXE_MODEL/LUXE_PROVIDER, then what you last picked, then default_model in models.json — and nothing under that, because Luxe never picks a model for you. Set false to always start from the configured default."
+ "description": "Remember the model and thinking level picked in a session and use them next launch. Precedence: --model/--thinking flag, then LUXE_MODEL/LUXE_PROVIDER, then what you last picked, then default_model in models.json — and nothing under that, because luxe never picks a model for you. Set false to always start from the configured default."
},
"on_ask": {
"type": "string",
@@ -195,7 +195,7 @@
"type": "string",
"enum": ["off", "streaming", "idle"],
"default": "streaming",
- "description": "When Luxe keeps a conversation's prompt cache alive through a long wait, by re-sending the last request with a one-token cap just before the cache entry expires (five minutes on Anthropic, unless supports_long_cache_retention asks for an hour). A refresh is a cache read of the prompt; the write it saves is usually ten times that. streaming (the default) does it while a turn is running — a slow tool, a permission prompt, a question — where the next request is certain. idle also does it for up to 30 minutes after a turn ends, where a reply is a guess, so a refresh is sent only on a conversation large enough to make the guess pay. off never does. Either way nothing is sent unless it is expected to save at least $0.05, never on a subscription plan, and its cost shows on /cost as cache warming. A project's config may turn this down and may not turn it up."
+ "description": "When luxe keeps a conversation's prompt cache alive through a long wait, by re-sending the last request with a one-token cap just before the cache entry expires (five minutes on Anthropic, unless supports_long_cache_retention asks for an hour). A refresh is a cache read of the prompt; the write it saves is usually ten times that. streaming (the default) does it while a turn is running — a slow tool, a permission prompt, a question — where the next request is certain. idle also does it for up to 30 minutes after a turn ends, where a reply is a guess, so a refresh is sent only on a conversation large enough to make the guess pay. off never does. Either way nothing is sent unless it is expected to save at least $0.05, never on a subscription plan, and its cost shows on /cost as cache warming. A project's config may turn this down and may not turn it up."
},
"default_mode": {
"type": "string",
@@ -289,7 +289,7 @@
},
"launchers": {
"type": "array",
- "description": "Programs a key hands the terminal to — gh-dash, lazygit, anything that takes over a terminal and gives it back. Luxe keeps running underneath, a turn included, and redraws when the program exits; text the program writes to the file named by $LUXE_HANDOFF lands in the composer. Read from the global config only: a key that runs a program a repository named is arbitrary execution.",
+ "description": "Programs a key hands the terminal to — gh-dash, lazygit, anything that takes over a terminal and gives it back. luxe keeps running underneath, a turn included, and redraws when the program exits; text the program writes to the file named by $LUXE_HANDOFF lands in the composer. Read from the global config only: a key that runs a program a repository named is arbitrary execution.",
"items": {
"type": "object",
"properties": {
@@ -319,7 +319,7 @@
},
"github": {
"type": "object",
- "description": "How Luxe keeps up with GitHub: the checked-out branch's pull request, and what /github lists. The refresh interval is read from the global config only — how often Luxe calls out on a timer is not a decision a repository you cloned makes; a project may add its own /github sections.",
+ "description": "How luxe keeps up with GitHub: the checked-out branch's pull request, and what /github lists. The refresh interval is read from the global config only — how often luxe calls out on a timer is not a decision a repository you cloned makes; a project may add its own /github sections.",
"properties": {
"refresh_secs": {
"type": "integer",
@@ -345,12 +345,12 @@
},
"update": {
"type": "object",
- "description": "Whether Luxe checks for a newer release. Read from the global config only — whether Luxe reaches the network on your behalf is not a decision a repository you cloned makes.",
+ "description": "Whether luxe checks for a newer release. Read from the global config only — whether luxe reaches the network on your behalf is not a decision a repository you cloned makes.",
"properties": {
"check": {
"type": "boolean",
"default": true,
- "description": "Look for a newer release, at most once a day, and mention it under the dashboard's footer. Luxe never updates itself; the notice names the command that matches how you installed it. The request sends nothing about you — no version, no platform, no identifier — and every failure is silent. Set false, or LUXE_UPDATE_CHECK=0, to turn it off."
+ "description": "Look for a newer release, at most once a day, and mention it under the dashboard's footer. luxe never updates itself; the notice names the command that matches how you installed it. The request sends nothing about you — no version, no platform, no identifier — and every failure is silent. Set false, or LUXE_UPDATE_CHECK=0, to turn it off."
}
},
"additionalProperties": false
diff --git a/crates/luxe-core/src/config/display.rs b/crates/luxe-core/src/config/display.rs
index 4d034d8..5073830 100644
--- a/crates/luxe-core/src/config/display.rs
+++ b/crates/luxe-core/src/config/display.rs
@@ -283,7 +283,7 @@ pub enum NotificationMethod {
/// The moments that send a desktop notification: all of them, none, or a list.
///
/// Read leniently, like every other key: a value that is neither a boolean nor a list is treated as
-/// unset and a name Luxe does not know is skipped, so one typo never takes the whole config with it.
+/// unset and a name luxe does not know is skipped, so one typo never takes the whole config with it.
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
#[serde(untagged)]
pub enum Notifications {
diff --git a/crates/luxe-core/src/config/display/diff.rs b/crates/luxe-core/src/config/display/diff.rs
index f78b43f..c43011b 100644
--- a/crates/luxe-core/src/config/display/diff.rs
+++ b/crates/luxe-core/src/config/display/diff.rs
@@ -3,7 +3,7 @@
use super::{known_or, Lenient};
use serde::{Deserialize, Serialize};
-/// How a diff is drawn, wherever Luxe draws one.
+/// How a diff is drawn, wherever luxe draws one.
///
/// One setting for three surfaces — the `edit`/`write` tool card, the peeked file beside the
/// conversation, and Guided Review. They differ only in how much room they have, and
diff --git a/crates/luxe-core/src/config/integrations.rs b/crates/luxe-core/src/config/integrations.rs
index bf75797..8d86de7 100644
--- a/crates/luxe-core/src/config/integrations.rs
+++ b/crates/luxe-core/src/config/integrations.rs
@@ -71,7 +71,7 @@ impl MentionSource {
!kind.is_empty() && !self.reserved() && !self.command.trim().is_empty()
}
- /// Whether its kind is one Luxe's own tabs use, which leaves it without a tab of its own.
+ /// Whether its kind is one luxe's own tabs use, which leaves it without a tab of its own.
///
/// Worth telling the reader rather than only skipping: `mcp` joined the list after sources
/// could be named anything, so a source that worked yesterday can be the one it displaced.
@@ -80,7 +80,7 @@ impl MentionSource {
}
}
-/// The reference kinds Luxe's own `@` tabs use: `@[file:…]`, `@[skill:…]`, `@[github:…]`,
+/// The reference kinds luxe's own `@` tabs use: `@[file:…]`, `@[skill:…]`, `@[github:…]`,
/// `@[mcp:…]`.
pub const RESERVED_MENTION_KINDS: [&str; 4] = ["file", "skill", "github", "mcp"];
@@ -102,7 +102,7 @@ impl WorktreeConfig {
/// GitHub preferences.
///
-/// Read from the global config only, like [`UpdateConfig`]: how often Luxe calls out on a timer is
+/// Read from the global config only, like [`UpdateConfig`]: how often luxe calls out on a timer is
/// the reader's decision, not one a repository they cloned gets to make.
#[derive(Debug, Clone, Serialize, Deserialize, Default, PartialEq)]
#[serde(default)]
@@ -152,7 +152,7 @@ impl GithubConfig {
/// Update-notice preferences.
///
-/// Luxe never updates itself; the only question here is whether it may *notice* that it is behind.
+/// luxe never updates itself; the only question here is whether it may *notice* that it is behind.
/// The check fetches a public manifest and sends nothing about you — see `luxe_core::update` — but
/// it is still an outbound request, so it gets a switch like every other one.
#[derive(Debug, Clone, Serialize, Deserialize, Default, PartialEq)]
@@ -186,7 +186,7 @@ pub struct LspConfig {
pub project_servers: Option,
}
-/// Whether Luxe may launch a language server out of the project's `node_modules/.bin`.
+/// Whether luxe may launch a language server out of the project's `node_modules/.bin`.
///
/// npm puts a dev dependency there rather than on `PATH`, so for the JavaScript servers it is
/// routinely the only copy that matches the project — `svelteserver` has to agree with the Svelte
@@ -223,7 +223,7 @@ pub struct ReviewConfig {
/// `PLAN.md`/`AGENTS.md`. Each must resolve inside the opened project.
#[serde(default)]
pub context_files: Vec,
- /// Project-relative globs a review leaves out, on top of Luxe's own `.luxe/` directory.
+ /// Project-relative globs a review leaves out, on top of luxe's own `.luxe/` directory.
///
/// For bookkeeping a repo carries but nobody reviews — a loop's `PROGRESS.md`, a planning
/// directory, another agent's memory. Excluded files are *counted and named*, never silently
diff --git a/crates/luxe-core/src/config/merge.rs b/crates/luxe-core/src/config/merge.rs
index 23ba4b8..e9126de 100644
--- a/crates/luxe-core/src/config/merge.rs
+++ b/crates/luxe-core/src/config/merge.rs
@@ -259,7 +259,7 @@ impl Layered for LspConfig {
}
impl Layered for UpdateConfig {
- // `global` only, same reasoning in a lower key: whether Luxe reaches the network on your
+ // `global` only, same reasoning in a lower key: whether luxe reaches the network on your
// behalf is your decision, not a decision a repository you cloned makes.
fn layered(global: Self, project: Self) -> Self {
let Self { check: _ } = project;
diff --git a/crates/luxe-core/src/config/tests.rs b/crates/luxe-core/src/config/tests.rs
index cd8d7cf..c5f24b8 100644
--- a/crates/luxe-core/src/config/tests.rs
+++ b/crates/luxe-core/src/config/tests.rs
@@ -98,7 +98,7 @@ fn partial_config_round_trips() {
assert_eq!(config.dock, DockConfig::default());
}
-/// Every leniently read setting, through its resolver: each value it knows, one from a newer Luxe,
+/// Every leniently read setting, through its resolver: each value it knows, one from a newer luxe,
/// and none at all.
#[test]
fn every_lenient_setting_resolves_known_unknown_and_absent_values() {
diff --git a/crates/luxe-core/src/config/tests/schema.rs b/crates/luxe-core/src/config/tests/schema.rs
index a352b5a..872f196 100644
--- a/crates/luxe-core/src/config/tests/schema.rs
+++ b/crates/luxe-core/src/config/tests/schema.rs
@@ -5,7 +5,7 @@ use super::*;
#[test]
fn schema_is_valid_json() {
let schema: serde_json::Value = serde_json::from_str(Config::schema_json()).unwrap();
- assert_eq!(schema["title"], "Luxe config");
+ assert_eq!(schema["title"], "luxe config");
assert!(schema.get("properties").is_some());
assert!(schema["properties"].get("$schema").is_some());
assert!(schema["properties"].get("notifications").is_some());
diff --git a/crates/luxe-core/src/config/write.rs b/crates/luxe-core/src/config/write.rs
index add579e..c7fe647 100644
--- a/crates/luxe-core/src/config/write.rs
+++ b/crates/luxe-core/src/config/write.rs
@@ -138,7 +138,7 @@ impl Config {
}
/// Edit the global config under [`JsonFile`]'s lock. Called from the UI thread; the wait for the
-/// lock is bounded (a few seconds), and only another Luxe writing the same file at that moment, or
+/// lock is bounded (a few seconds), and only another luxe writing the same file at that moment, or
/// one that crashed holding the lock, makes it wait at all.
fn edit_global(edit: impl FnOnce(&mut Map)) -> Result {
let path = Config::global_path();
diff --git a/crates/luxe-core/src/context.rs b/crates/luxe-core/src/context.rs
index 49bb047..3122dd0 100644
--- a/crates/luxe-core/src/context.rs
+++ b/crates/luxe-core/src/context.rs
@@ -27,10 +27,10 @@ use std::path::PathBuf;
#[cfg(test)]
const HARNESS_TOOLS: &[&str] = &["question", "todo"];
-/// Where questions about Luxe itself are answered: its own guide, built into the binary and run
+/// Where questions about luxe itself are answered: its own guide, built into the binary and run
/// through `bash`. Static text, so the prompt stays byte-identical between turns; without it, a
-/// model asked how to configure Luxe answers from what other harnesses do.
-const LUXE_DOCS: &str = " Questions about Luxe itself are answered by its guide: `luxe docs`.";
+/// model asked how to configure luxe answers from what other harnesses do.
+const LUXE_DOCS: &str = " Questions about luxe itself are answered by its guide: `luxe docs`.";
/// Purpose-built tools paired with the shell command a model reaches for instead.
///
@@ -239,7 +239,7 @@ impl SystemPromptBuilder {
format!(
"## Tools\n\
- Every tool in Luxe is designed to beat the shell command it replaces by providing \
+ Every tool in luxe is designed to beat the shell command it replaces by providing \
structured, bounded output for a fraction of the context. They also make your actions \
easier for the user to read, which improves clarity and efficiency. So, always reach \
for the purpose-built tool before `bash`, which should be reserved for everything they \
@@ -294,7 +294,7 @@ pub(crate) fn shared_len(sections: &[MeasuredSection]) -> usize {
}
/// The `## Memory` section: the repository as what outlives the conversation, and where to ask
-/// about Luxe itself when the agent can run `luxe docs`.
+/// about luxe itself when the agent can run `luxe docs`.
fn memory_section(ctx: &PromptContext) -> String {
let docs = if ctx.tools.iter().any(|tool| tool == "bash") { LUXE_DOCS } else { "" };
format!(
@@ -335,7 +335,7 @@ fn plans_section() -> String {
/// Who the agent is, in every session and every mode.
pub(crate) const DEFAULT_ROLE: &str =
- "You are an agentic engineering assistant operating in Luxe, a coding \
+ "You are an agentic engineering assistant operating in luxe, a coding \
agent harness. Build features incrementally with small, safe diffs, and verify them before calling \
the work done.";
@@ -382,7 +382,7 @@ own conventions say otherwise:
- Comments should say *why*, never *what*.
- Use specific names and types: no `data`, `handler`, or `Manager`.
- Errors should be detailed and never swallowed.
-- Luxe features specialized tools to read, write, and debug code. Use them.";
+- luxe features specialized tools to read, write, and debug code. Use them.";
/// What makes a test worth having.
const TESTING: &str = "## Testing
diff --git a/crates/luxe-core/src/context/tests.rs b/crates/luxe-core/src/context/tests.rs
index 3e2b170..34c6904 100644
--- a/crates/luxe-core/src/context/tests.rs
+++ b/crates/luxe-core/src/context/tests.rs
@@ -26,7 +26,7 @@ fn ctx() -> PromptContext {
}
}
-/// An agent that can run commands is told where Luxe's own guide is, so a question about Luxe
+/// An agent that can run commands is told where luxe's own guide is, so a question about luxe
/// is answered from it; one that cannot run `luxe docs` is not told about it.
#[test]
fn the_prompt_points_at_luxes_own_guide() {
@@ -122,7 +122,7 @@ fn the_harness_names_no_tool_it_cannot_offer() {
/// It is product, not taste. A model asked for something it cannot determine will reach for a
/// plausible answer instead of admitting the gap, and a fabricated one is worse than a refusal
/// precisely because nothing distinguishes it from a real one — so each clause here is a
-/// specific shape that failure takes, and losing any of them silently makes Luxe less
+/// specific shape that failure takes, and losing any of them silently makes luxe less
/// trustworthy rather than merely less tidy.
#[test]
fn the_honesty_contract_is_always_on() {
@@ -285,7 +285,7 @@ fn no_live_work_state_reaches_the_cached_prefix() {
/// It is product, not taste: "write a test" is an instruction a model already follows, and
/// following only that is what produced the hollow suites this section exists to prevent. Each
/// clause below is one of those failure shapes, so losing any of them silently is a
-/// regression in what Luxe writes rather than in how it reads.
+/// regression in what luxe writes rather than in how it reads.
#[test]
fn the_testing_contract_is_always_on() {
{
diff --git a/crates/luxe-core/src/context_report.rs b/crates/luxe-core/src/context_report.rs
index 1106486..6d2c7a8 100644
--- a/crates/luxe-core/src/context_report.rs
+++ b/crates/luxe-core/src/context_report.rs
@@ -187,7 +187,7 @@ pub struct ContextComposition {
pub cache_write_pct: Option,
/// Enabled MCP servers, and the tokens their single gateway tool costs.
///
- /// A line rather than a part: Luxe exposes every server through one gateway tool, so the
+ /// A line rather than a part: luxe exposes every server through one gateway tool, so the
/// category the rest of the field warns about is always near zero here — and a row that always
/// reads zero teaches nothing.
pub mcp: Option,
diff --git a/crates/luxe-core/src/engine.rs b/crates/luxe-core/src/engine.rs
index fad44b3..def8c8f 100644
--- a/crates/luxe-core/src/engine.rs
+++ b/crates/luxe-core/src/engine.rs
@@ -81,7 +81,7 @@ pub struct Engine {
/// Mutable session id — updated when loading/resuming a session so the next turn chains
/// from the resumed state, matching the `--resume` path.
session_id: tokio::sync::Mutex,
- /// This process's claim on `session_id`, which keeps a second Luxe from writing it too
+ /// This process's claim on `session_id`, which keeps a second luxe from writing it too
session_claim: std::sync::Mutex>,
last_turn_baseline: tokio::sync::Mutex >,
active_guided_review: Arc>,
diff --git a/crates/luxe-core/src/engine/answering.rs b/crates/luxe-core/src/engine/answering.rs
index 44604b1..367743a 100644
--- a/crates/luxe-core/src/engine/answering.rs
+++ b/crates/luxe-core/src/engine/answering.rs
@@ -120,7 +120,7 @@ fn threads_of(
impl Engine {
/// Reports what in the working tree is not committed, asked fresh from git rather than read
- /// from the last snapshot, which may predate a file made outside Luxe.
+ /// from the last snapshot, which may predate a file made outside luxe.
pub(crate) fn check_worktree_removal(&self, event_tx: &mpsc::Sender) {
let root = self.config.guided_review_project_root.clone();
let event_tx = event_tx.clone();
diff --git a/crates/luxe-core/src/engine/config.rs b/crates/luxe-core/src/engine/config.rs
index 2848bc6..efeb085 100644
--- a/crates/luxe-core/src/engine/config.rs
+++ b/crates/luxe-core/src/engine/config.rs
@@ -53,7 +53,7 @@ pub struct EngineConfig {
///
/// `None` disables runtime review loading for embedding and test callers.
pub guided_review_project_root: Option,
- /// The directory Luxe was launched in, when that is not the project root.
+ /// The directory luxe was launched in, when that is not the project root.
///
/// The one thing the *shell's* directory still answers after [`crate::project::discover`] has
/// moved the process to the top of the working copy: which instruction files apply. A monorepo
@@ -72,9 +72,13 @@ pub struct EngineConfig {
pub review_context_files: Vec,
/// Project-relative globs a review leaves out (`review.exclude`), on top of `.luxe/`.
pub review_exclude: Vec,
- /// Whether an in-session model/thinking choice is remembered for the next launch
- /// (`agent.remember_model`). Off for embedders and tests, which should leave no trace.
- pub remember_runtime_state: bool,
+ /// The project an in-session model or thinking choice is remembered for, for the next launch
+ /// (`agent.remember_model`).
+ ///
+ /// A model is kept for this project and as the latest anywhere, so the next launch here starts
+ /// on it and a project never opened before starts on it too. `None` for embedders and tests,
+ /// which should leave no trace.
+ pub remember_choices_in: Option,
/// When the prompt cache is kept warm through a long wait (`agent.cache_warming`).
///
/// Only matters for a provider whose `prompt_cache_ttl` answers, so an embedder's or a test's
diff --git a/crates/luxe-core/src/engine/dispatch/permission.rs b/crates/luxe-core/src/engine/dispatch/permission.rs
index 5f4e4f6..54503bc 100644
--- a/crates/luxe-core/src/engine/dispatch/permission.rs
+++ b/crates/luxe-core/src/engine/dispatch/permission.rs
@@ -223,7 +223,7 @@ impl Engine {
/// or stopping. What it should do is hand the command to the person.
pub(super) fn deny_message(request: &crate::permission::PermissionRequest) -> String {
if crate::permission::defaults::runs_as_root(request) {
- return "Blocked: Luxe does not run commands as root. A command that needs root needs the \
+ return "Blocked: luxe does not run commands as root. A command that needs root needs the \
user's attention and consent, so they run it: give them the exact command in a \
fenced code block, ask them to run it in their own terminal, and continue once \
they say it is done. Do not look for another way to run it here."
diff --git a/crates/luxe-core/src/engine/dispatch/test_nudge.rs b/crates/luxe-core/src/engine/dispatch/test_nudge.rs
index 4b02560..bfa5b1e 100644
--- a/crates/luxe-core/src/engine/dispatch/test_nudge.rs
+++ b/crates/luxe-core/src/engine/dispatch/test_nudge.rs
@@ -4,7 +4,7 @@ use super::*;
impl Engine {
/// Nudge toward the `debug` tool when tests keep failing — runtime reinforcement for models
- /// that default to print-and-retry (the debugger is a Luxe differentiator, but under-used).
+ /// that default to print-and-retry (the debugger is a luxe differentiator, but under-used).
/// Fires once per failing streak (≥3 consecutive failing test runs); a passing test resets it.
/// Disable with `LUXE_DEBUG_NUDGE=0`.
pub(super) async fn maybe_nudge_debugger(&self, is_test: bool, result: &mut ToolResultData) {
diff --git a/crates/luxe-core/src/engine/drafting.rs b/crates/luxe-core/src/engine/drafting.rs
index c3476d8..833985a 100644
--- a/crates/luxe-core/src/engine/drafting.rs
+++ b/crates/luxe-core/src/engine/drafting.rs
@@ -2,7 +2,7 @@
//!
//! A draft is an ordinary turn in this session, so it is written by the agent that did the work,
//! with the transcript it already holds in the prompt cache. What the turn adds is appended once,
-//! as one message, never in the system prompt: the thing being answered in full, Luxe's own evidence
+//! as one message, never in the system prompt: the thing being answered in full, luxe's own evidence
//! about it — the code at the line now, the commits since, the review comment it became and whether
//! the fix was verified — and the rest of the conversation as a digest, sent whole the first time
//! and only what is new after, until compaction or a session load forgets what was sent.
diff --git a/crates/luxe-core/src/engine/drafting/tests.rs b/crates/luxe-core/src/engine/drafting/tests.rs
index ff30807..b658c52 100644
--- a/crates/luxe-core/src/engine/drafting/tests.rs
+++ b/crates/luxe-core/src/engine/drafting/tests.rs
@@ -155,7 +155,7 @@ async fn a_thread_brief_carries_the_code_now_and_the_commits_since() {
let dir = tempfile::tempdir().unwrap();
let root = dir.path();
git(root, &["init", "--quiet", "-b", "main"]);
- git(root, &["config", "user.name", "Luxe Tests"]);
+ git(root, &["config", "user.name", "luxe Tests"]);
git(root, &["config", "user.email", "tests@luxe.local"]);
let lines: Vec = (1..=20).map(|n| format!("line {n}")).collect();
std::fs::write(root.join("router.rs"), lines.join("\n")).unwrap();
diff --git a/crates/luxe-core/src/engine/error.rs b/crates/luxe-core/src/engine/error.rs
index a1c391c..ee30d8c 100644
--- a/crates/luxe-core/src/engine/error.rs
+++ b/crates/luxe-core/src/engine/error.rs
@@ -22,7 +22,7 @@ pub enum EngineError {
///
/// The gate, rather than the courtesy: a front-end should ask before it submits, and this is
/// what holds when one does not. See [`crate::provider::ModelRef::is_unset`].
- #[error("no model selected — pick one with /model, or start Luxe with --model ")]
+ #[error("no model selected — pick one with /model, or start luxe with --model ")]
NoModel,
}
@@ -48,7 +48,7 @@ mod tests {
let session = SessionError::Other("gone".into());
let session_said = format!("session error: {session}");
let no_model =
- "no model selected — pick one with /model, or start Luxe with --model ";
+ "no model selected — pick one with /model, or start luxe with --model ";
let cases: [(EngineError, String, bool); 5] = [
(provider.into(), provider_said, false),
(session.into(), session_said, false),
diff --git a/crates/luxe-core/src/engine/github_overview.rs b/crates/luxe-core/src/engine/github_overview.rs
index 7259fb3..a4d42e8 100644
--- a/crates/luxe-core/src/engine/github_overview.rs
+++ b/crates/luxe-core/src/engine/github_overview.rs
@@ -5,7 +5,7 @@
//! repository and number asked about, so a reply that lands after the reader moved on is matched
//! to nothing rather than drawn over the wrong row.
//!
-//! An issue's answer also says whether one of your sessions took it up — Luxe's own knowledge, read
+//! An issue's answer also says whether one of your sessions took it up — luxe's own knowledge, read
//! from the session store, which GitHub has no way to know.
use super::posting::sentence;
diff --git a/crates/luxe-core/src/engine/guided_review/baseline.rs b/crates/luxe-core/src/engine/guided_review/baseline.rs
index 88fc90b..5fe8730 100644
--- a/crates/luxe-core/src/engine/guided_review/baseline.rs
+++ b/crates/luxe-core/src/engine/guided_review/baseline.rs
@@ -41,7 +41,7 @@ impl Engine {
}
}
- /// What this engine leaves out of a review: Luxe's own directory, plus `review.exclude`.
+ /// What this engine leaves out of a review: luxe's own directory, plus `review.exclude`.
pub(crate) fn review_exclusions(&self) -> crate::guided_review::ReviewExclusions {
crate::guided_review::ReviewExclusions::new(&self.config.review_exclude)
}
diff --git a/crates/luxe-core/src/engine/guided_review/target.rs b/crates/luxe-core/src/engine/guided_review/target.rs
index 473e22a..fb3b927 100644
--- a/crates/luxe-core/src/engine/guided_review/target.rs
+++ b/crates/luxe-core/src/engine/guided_review/target.rs
@@ -43,7 +43,7 @@ pub(super) fn pull_request_failure(error: &PullRequestResolveError) -> GuidedRev
GuidedReviewFailure::unavailable(format!("Could not run `gh`: {source}."))
}
GitHubError::Unreadable(_) => GuidedReviewFailure::unavailable(
- "`gh` answered with something Luxe could not read. Updating the GitHub CLI may fix it.",
+ "`gh` answered with something luxe could not read. Updating the GitHub CLI may fix it.",
),
GitHubError::RateLimited => GuidedReviewFailure::unavailable(
"GitHub's rate limit for this account is spent. It refills within the hour.",
diff --git a/crates/luxe-core/src/engine/lifecycle.rs b/crates/luxe-core/src/engine/lifecycle.rs
index 98d7d9c..001efaf 100644
--- a/crates/luxe-core/src/engine/lifecycle.rs
+++ b/crates/luxe-core/src/engine/lifecycle.rs
@@ -190,7 +190,7 @@ impl Engine {
.read()
.map_err(|_| SessionError::Other("session store lock poisoned".to_string()))?;
let entries = store.load(&id)?;
- // Before anything changes, so a session open in another Luxe leaves this one as it was
+ // Before anything changes, so a session open in another luxe leaves this one as it was
let claim = store.claim(&id)?;
let leaf = crate::session::latest_leaf(&entries);
let active = match leaf.as_ref() {
diff --git a/crates/luxe-core/src/engine/nested_instructions.rs b/crates/luxe-core/src/engine/nested_instructions.rs
index 9864a15..0a51670 100644
--- a/crates/luxe-core/src/engine/nested_instructions.rs
+++ b/crates/luxe-core/src/engine/nested_instructions.rs
@@ -74,7 +74,7 @@ impl Engine {
}
}
-/// The nested files `call` reached, from Luxe launched at `launch`.
+/// The nested files `call` reached, from luxe launched at `launch`.
///
/// Relative paths resolve the way the tool resolved them, against where tools run.
fn reached(launch: &std::path::Path, call: &ToolCallInfo) -> Vec {
diff --git a/crates/luxe-core/src/engine/posting.rs b/crates/luxe-core/src/engine/posting.rs
index ec3fa45..0def27e 100644
--- a/crates/luxe-core/src/engine/posting.rs
+++ b/crates/luxe-core/src/engine/posting.rs
@@ -267,7 +267,7 @@ pub(super) fn sentence(error: &GitHubError) -> String {
}
GitHubError::Run { source, .. } => format!("Could not run `gh`: {source}."),
GitHubError::Unreadable(_) => {
- "`gh` answered with something Luxe could not read. Updating the GitHub CLI may fix it."
+ "`gh` answered with something luxe could not read. Updating the GitHub CLI may fix it."
.into()
}
GitHubError::RateLimited => {
diff --git a/crates/luxe-core/src/engine/prompt.rs b/crates/luxe-core/src/engine/prompt.rs
index 9925fe3..21b8c0c 100644
--- a/crates/luxe-core/src/engine/prompt.rs
+++ b/crates/luxe-core/src/engine/prompt.rs
@@ -12,7 +12,7 @@ impl Engine {
pub(super) async fn assemble_prompt(
&self,
) -> Result<(Vec, Option), EngineError> {
- // Two directories, and they differ whenever Luxe is launched below the project root: the
+ // Two directories, and they differ whenever luxe is launched below the project root: the
// one the launch named, which only decides which `AGENTS.md` files apply, and the one every
// tool actually runs in. The prompt reports the second as `cwd`: a session opened in
// `packages/app` must not be told it is there while `pwd` answers with the root, and a
@@ -147,7 +147,7 @@ fn build_deferred_tools_section(catalog: &[(String, String)]) -> String {
out
}
-/// Hand-written catalog lines for Luxe's own deferred tools, shorter than anything
+/// Hand-written catalog lines for luxe's own deferred tools, shorter than anything
/// [`catalog_line`] can cut from a description. A tool not listed here, an MCP server's or a newer
/// one, gets the generated line.
const CATALOG_SUMMARIES: &[(&str, &str)] = &[
diff --git a/crates/luxe-core/src/engine/review_desk/publishing.rs b/crates/luxe-core/src/engine/review_desk/publishing.rs
index ed0a95d..3b12b5d 100644
--- a/crates/luxe-core/src/engine/review_desk/publishing.rs
+++ b/crates/luxe-core/src/engine/review_desk/publishing.rs
@@ -69,7 +69,7 @@ impl ReviewDesk {
}
if interrupted > 0 {
effects.push(Effect::Event(Event::Notice(
- "A review feedback turn was cut off when Luxe last stopped. It can be sent again \
+ "A review feedback turn was cut off when luxe last stopped. It can be sent again \
from the review."
.into(),
)));
diff --git a/crates/luxe-core/src/engine/review_desk/storage.rs b/crates/luxe-core/src/engine/review_desk/storage.rs
index 0365c2d..ef6cb9c 100644
--- a/crates/luxe-core/src/engine/review_desk/storage.rs
+++ b/crates/luxe-core/src/engine/review_desk/storage.rs
@@ -126,13 +126,13 @@ impl ReviewDesk {
"Review comments cannot be saved without a project.".to_string()
}
SaveError::Store(ReviewSessionStoreError::RevisionConflict { .. }) => {
- // Another Luxe on this project wrote first. Holding on to the older revision would
+ // Another luxe on this project wrote first. Holding on to the older revision would
// refuse every later change too, so take theirs and say so.
if let Some(stored) = self.load(&review_id).await {
self.session = Some(stored);
self.persisted = true;
}
- "Another Luxe changed this review's comments at the same time, so this change was \
+ "Another luxe changed this review's comments at the same time, so this change was \
not saved. The review shows theirs now; make the change again."
.to_string()
}
diff --git a/crates/luxe-core/src/engine/review_desk/tests/edits.rs b/crates/luxe-core/src/engine/review_desk/tests/edits.rs
index f8d61fd..ddf602c 100644
--- a/crates/luxe-core/src/engine/review_desk/tests/edits.rs
+++ b/crates/luxe-core/src/engine/review_desk/tests/edits.rs
@@ -245,7 +245,7 @@ async fn carried_comments_join_and_only_the_ones_still_placed_are_sent() {
assert!(submitted(&effects).is_some());
}
-/// A revision conflict (another Luxe wrote first) refuses the edit, takes the stored session,
+/// A revision conflict (another luxe wrote first) refuses the edit, takes the stored session,
/// and lets the next edit through.
///
/// The break this catches: holding on to the older revision, which refuses every edit after it too.
@@ -261,7 +261,7 @@ async fn a_revision_conflict_takes_the_stored_session_and_recovers() {
let (request_id, effects) =
edit(&mut desk, ReviewEdit::SetOutcome { outcome: ReviewOutcome::Accepted }).await;
- assert!(refusal_for(&effects, &request_id).contains("Another Luxe"));
+ assert!(refusal_for(&effects, &request_id).contains("Another luxe"));
let restored = effects.iter().find_map(|effect| match effect {
Effect::Event(Event::ReviewSessionRestored { session }) => Some(session.outcome),
_ => None,
diff --git a/crates/luxe-core/src/engine/settings.rs b/crates/luxe-core/src/engine/settings.rs
index 44a2540..cecdc7f 100644
--- a/crates/luxe-core/src/engine/settings.rs
+++ b/crates/luxe-core/src/engine/settings.rs
@@ -195,8 +195,9 @@ impl Engine {
event_tx: &mpsc::Sender,
) {
debug!(model = %model.r#ref, "switching active model");
- if self.config.remember_runtime_state {
+ if let Some(project) = &self.config.remember_choices_in {
crate::runtime_state::RuntimeState::remember_model(
+ project,
&model.r#ref.provider,
&model.r#ref.id,
);
@@ -217,7 +218,7 @@ impl Engine {
event_tx: &mpsc::Sender,
) {
debug!(?level, "setting thinking level");
- if self.config.remember_runtime_state {
+ if self.config.remember_choices_in.is_some() {
crate::runtime_state::RuntimeState::remember_thinking(level);
}
*self.before_mode.lock().await = None;
diff --git a/crates/luxe-core/src/engine/turn.rs b/crates/luxe-core/src/engine/turn.rs
index 9049872..d16caba 100644
--- a/crates/luxe-core/src/engine/turn.rs
+++ b/crates/luxe-core/src/engine/turn.rs
@@ -292,7 +292,7 @@ impl Engine {
*self.draft_answer.lock().await = committed.text.clone();
}
// The checkpoint that delivers a steer only runs at the top of an iteration, so ending here
- // would leave it buffered until the user's next submission — a message they watched Luxe
+ // would leave it buffered until the user's next submission — a message they watched luxe
// accept and then apparently ignore. Each continuation consumes the buffer, so this can
// only extend the turn as often as the user actually types.
if !self.pending_steers.lock().await.is_empty() {
diff --git a/crates/luxe-core/src/engine/turn/commit.rs b/crates/luxe-core/src/engine/turn/commit.rs
index 5ea4a65..4fabab5 100644
--- a/crates/luxe-core/src/engine/turn/commit.rs
+++ b/crates/luxe-core/src/engine/turn/commit.rs
@@ -33,7 +33,7 @@ impl Engine {
///
/// Every committed message is announced with `MessageCommitted`, the reader's own words too:
/// a client joined to the session didn't type them, and the front-end that did reconciles the
- /// copy it already shows. A message Luxe composed (review feedback, a draft's brief) has no
+ /// copy it already shows. A message luxe composed (review feedback, a draft's brief) has no
/// such copy anywhere, so it's announced the same way.
pub(super) async fn commit_user_message(
&self,
@@ -142,7 +142,7 @@ impl Engine {
/// Reports a provider call that came back with nothing, and says why the turn stops.
///
/// A cancel is the user's own doing and the UI has already said so; anything else is the
- /// model answering with silence, which looked exactly like Luxe hanging — the spinner
+ /// model answering with silence, which looked exactly like luxe hanging — the spinner
/// stopped, no message appeared, and the next thing the reader saw was a `400` from the turn
/// after it.
pub(super) async fn report_empty_reply(&self, event_tx: &mpsc::Sender) -> StopReason {
diff --git a/crates/luxe-core/src/engine/turn/report.rs b/crates/luxe-core/src/engine/turn/report.rs
index c3102c9..c8fc618 100644
--- a/crates/luxe-core/src/engine/turn/report.rs
+++ b/crates/luxe-core/src/engine/turn/report.rs
@@ -2,7 +2,7 @@
//!
//! A fix request ends with the report the agent is to fill in. The turn reads everything the agent
//! said in it, every message rather than only the last, as the turn is about to end. A report that
-//! reads goes to the review. When there is none Luxe can use, Luxe appends one message saying what
+//! reads goes to the review. When there is none luxe can use, luxe appends one message saying what
//! was wrong, with the template again, and the turn goes on for the answer: once, so a model that
//! cannot comply costs a call rather than a loop. Whatever still has no word from the agent when
//! the turn ends, however it ended, goes to the review desk, which notes it on the comment.
@@ -16,9 +16,9 @@ use crate::guided_review::FindingId;
pub(super) struct ReportWatch {
/// Every assistant text the turn committed, in order
said: Vec,
- /// Whether Luxe has asked for the report this turn
+ /// Whether luxe has asked for the report this turn
asked: bool,
- /// Why the last reading found no report Luxe could use
+ /// Why the last reading found no report luxe could use
problem: Option<&'static str>,
/// The comments the delivered report spoke for; `None` until one is delivered
covered: Option>,
@@ -141,7 +141,7 @@ const NO_REPORT: &str = "it wrote no report";
/// No `Submission:` line, which is how a retried submission finds the message it already sent.
fn report_request(context: &RemediationTurnContext, reason: &str) -> String {
format!(
- "{REPORT_REQUEST}\nYour reply did not give the report Luxe reads to update the review: \
+ "{REPORT_REQUEST}\nYour reply did not give the report luxe reads to update the review: \
{reason}. Do not redo any work. Reply with just this block, one line per \
comment in the order they were sent, each status set to fixed, not_fixed or \
needs_clarification and each note saying what you did, in one sentence.\n\n{}",
diff --git a/crates/luxe-core/src/engine/turn/request.rs b/crates/luxe-core/src/engine/turn/request.rs
index c996ebc..4ee35f0 100644
--- a/crates/luxe-core/src/engine/turn/request.rs
+++ b/crates/luxe-core/src/engine/turn/request.rs
@@ -133,7 +133,7 @@ impl Engine {
/// on the way in is an estimate — char/4 for anything not yet measured, a tokenizer we do not
/// have for the rest — so a prompt can clear the threshold and still be rejected, and a
/// mid-turn tool result large enough to do it arrives *after* the last check. Failing there
- /// ends the turn on the one error whose remedy Luxe already implements.
+ /// ends the turn on the one error whose remedy luxe already implements.
///
/// The retry summarizes on the provider's authority rather than the threshold's: the window is
/// full because the endpoint says so, whatever the local arithmetic believed. Once only, and
diff --git a/crates/luxe-core/src/event/action.rs b/crates/luxe-core/src/event/action.rs
index 9136e91..4c3e102 100644
--- a/crates/luxe-core/src/event/action.rs
+++ b/crates/luxe-core/src/event/action.rs
@@ -168,7 +168,7 @@ pub enum Action {
squash: bool,
},
/// Draft the pull request the open review would open, answered by `pull_request_prepared`.
- /// Luxe keeps the draft for `open_pull_request`.
+ /// luxe keeps the draft for `open_pull_request`.
PreparePullRequest {
/// Names the draft
request_id: crate::guided_review::PullRequestOpenId,
@@ -189,7 +189,7 @@ pub enum Action {
draft: bool,
},
/// Draft the open review as a stack, one pull request per chapter, answered by
- /// `stack_prepared`. Luxe keeps the layers for `open_stack`.
+ /// `stack_prepared`. luxe keeps the layers for `open_stack`.
PrepareStack {
/// Names the stack
request_id: crate::guided_review::PullRequestOpenId,
@@ -205,7 +205,7 @@ pub enum Action {
draft: bool,
},
/// Read what posting the open pull request review needs, answered by
- /// `pull_request_post_prepared`. Luxe keeps it for `post_pull_request_review`.
+ /// `pull_request_post_prepared`. luxe keeps it for `post_pull_request_review`.
PreparePullRequestPost {
/// Names the preview
request_id: crate::guided_review::PullRequestPostId,
@@ -293,7 +293,7 @@ pub enum Action {
repository: Option,
},
/// Work out what restacking the checked-out branch's stack would do, answered by
- /// `restack_planned`. Luxe keeps the plan for the `restack` that confirms it.
+ /// `restack_planned`. luxe keeps the plan for the `restack` that confirms it.
PlanRestack {
/// Names the plan
request_id: crate::restack::RestackId,
@@ -331,7 +331,7 @@ pub enum Action {
/// Project-relative file
path: String,
},
- /// Whether the client's window has focus, which paces Luxe's background GitHub refreshes.
+ /// Whether the client's window has focus, which paces luxe's background GitHub refreshes.
SetFocus {
/// Whether it now has focus
focused: bool,
diff --git a/crates/luxe-core/src/event/events.rs b/crates/luxe-core/src/event/events.rs
index 849acd5..69af54c 100644
--- a/crates/luxe-core/src/event/events.rs
+++ b/crates/luxe-core/src/event/events.rs
@@ -516,17 +516,17 @@ pub enum Event {
/// The answer to `show_open_review`. Boxed: it carries a whole review.
OpenReviewShown(Box),
/// The open review's feedback session changed and was stored. Every change arrives this way,
- /// whoever made it: an `EditReview`, answered under its `request_id`, or something Luxe did
+ /// whoever made it: an `EditReview`, answered under its `request_id`, or something luxe did
/// itself (a turn dispatched, a report recorded, a review posted), under a fresh one.
ReviewSessionSaved {
- /// The edit this answers, or a fresh id for a change Luxe made itself.
+ /// The edit this answers, or a fresh id for a change luxe made itself.
request_id: ReviewSessionSaveRequestId,
/// Complete authoritative state returned by the expected-revision store.
session: Box,
},
/// A change to the feedback session was refused or could not be stored, and nothing changed.
ReviewSessionSaveFailed {
- /// The edit this answers, or a fresh id for a change Luxe made itself.
+ /// The edit this answers, or a fresh id for a change luxe made itself.
request_id: ReviewSessionSaveRequestId,
/// Logical review whose mutable state was not persisted.
review_id: ReviewId,
@@ -586,7 +586,7 @@ pub enum Event {
Ready {
/// Wire protocol version — [`crate::PROTOCOL_VERSION`].
protocol_version: u32,
- /// The Luxe build serving this session.
+ /// The luxe build serving this session.
luxe_version: String,
/// Session the stream is attached to.
session_id: String,
diff --git a/crates/luxe-core/src/event/tests/decoding.rs b/crates/luxe-core/src/event/tests/decoding.rs
index 134648d..84fb6d8 100644
--- a/crates/luxe-core/src/event/tests/decoding.rs
+++ b/crates/luxe-core/src/event/tests/decoding.rs
@@ -6,7 +6,7 @@ use super::*;
/// read it before the derived ones replaced them.
///
/// The break this catches: the mirror enums decode only what their attributes say, so a
-/// `#[serde(default)]` missing there turns a frame a client or an older Luxe sends into an error.
+/// `#[serde(default)]` missing there turns a frame a client or an older luxe sends into an error.
/// Every row leaves out a field that is not an `Option`, which decodes as `None` regardless.
#[test]
fn what_a_sender_may_leave_out_still_decodes() {
diff --git a/crates/luxe-core/src/external_work.rs b/crates/luxe-core/src/external_work.rs
index bc897a2..e809cbd 100644
--- a/crates/luxe-core/src/external_work.rs
+++ b/crates/luxe-core/src/external_work.rs
@@ -11,7 +11,7 @@
//!
//! # What it can and cannot see
//!
-//! Only paths Luxe's own file tools name. A `bash` command that deletes a file elsewhere is
+//! Only paths luxe's own file tools name. A `bash` command that deletes a file elsewhere is
//! invisible here, which is why an external path inside a Git worktree is recorded as a *root*
//! instead — Git then reports that tree the same way it reports the project, and catches what the
//! shell did too. `ExternalWork::roots` is the honest answer for a checkout; `ExternalWork::loose`
@@ -129,7 +129,7 @@ impl ExternalWork {
/// Called *before* the change, so the first call for a path is the only one that can see what
/// was there; later calls leave the retained baseline alone. A path inside a Git worktree is
/// recorded as its root instead of as a file, because Git can then report that tree completely
- /// rather than only the parts Luxe touched.
+ /// rather than only the parts luxe touched.
pub fn observe(&self, path: &Path) {
let Ok(mut state) = self.state.lock() else {
return;
diff --git a/crates/luxe-core/src/fetched.rs b/crates/luxe-core/src/fetched.rs
index 83a28cc..06df9dc 100644
--- a/crates/luxe-core/src/fetched.rs
+++ b/crates/luxe-core/src/fetched.rs
@@ -2,7 +2,7 @@
//!
//! A clone lives inside the project so the agent can read and search it without tripping the
//! external-directory guard. That makes it the one directory in the tree whose contents came from a
-//! stranger rather than from the person running Luxe, so anything that would *run* repository code
+//! stranger rather than from the person running luxe, so anything that would *run* repository code
//! on the user's behalf — a language server rooted there, a formatter loading its plugins — checks
//! [`contains`] and stays out. This module is the single answer to where that directory is and
//! whether a path is inside it.
diff --git a/crates/luxe-core/src/file_diff.rs b/crates/luxe-core/src/file_diff.rs
index a10cb3c..00c06a8 100644
--- a/crates/luxe-core/src/file_diff.rs
+++ b/crates/luxe-core/src/file_diff.rs
@@ -6,7 +6,7 @@
//! This module answers that second question with one `git diff`, parsed by the review's own
//! [`parse_unified_diff`] into the review's own [`ReviewFile`], and nothing kept afterwards.
//!
-//! Reusing the review's content model is the point: there is one diff data model in Luxe, so the
+//! Reusing the review's content model is the point: there is one diff data model in luxe, so the
//! Changes pane and the review cannot disagree about what a hunk is, and escalating from one to
//! the other shows the same lines. What this does *not* share is the capture: no private index
//! snapshot, no stability re-read, no artifact on disk, no chaptering. A glance is allowed to be
@@ -110,7 +110,7 @@ pub enum FileDiffFailure {
#[error("this file is not text, so there are no lines to show")]
NotText,
/// Git produced a diff the review parser rejected.
- #[error("git produced a diff Luxe could not read: {0}")]
+ #[error("git produced a diff luxe could not read: {0}")]
Unreadable(String),
}
@@ -228,7 +228,7 @@ async fn capture_loose(
requested: &str,
baseline: Option<&ExternalBaseline>,
) -> Result {
- // A baseline of `Unretained` and no record at all are the same answer to the reader: Luxe
+ // A baseline of `Unretained` and no record at all are the same answer to the reader: luxe
// cannot say what this file used to hold, so it will not pretend to.
let Some(baseline @ (ExternalBaseline::Absent | ExternalBaseline::Text { .. })) = baseline
else {
diff --git a/crates/luxe-core/src/formatter.rs b/crates/luxe-core/src/formatter.rs
index 901fb16..9d7e689 100644
--- a/crates/luxe-core/src/formatter.rs
+++ b/crates/luxe-core/src/formatter.rs
@@ -289,7 +289,7 @@ pub async fn run_formatter(
/// looks — on a 2024 crate it formats to the wrong style edition and refuses to parse syntax that
/// edition added, which reads as a broken file rather than a wrong flag. `rustfmt.toml` first,
/// because that is the project speaking about formatting specifically; then the manifest; then the
-/// last edition Luxe can safely assume.
+/// last edition luxe can safely assume.
///
/// A line scan rather than a TOML parse, deliberately: one key, three well-known file names, and no
/// dependency worth adding for it. Anything it cannot read falls back rather than guessing.
diff --git a/crates/luxe-core/src/formatter/tests.rs b/crates/luxe-core/src/formatter/tests.rs
index 657af83..479ccbc 100644
--- a/crates/luxe-core/src/formatter/tests.rs
+++ b/crates/luxe-core/src/formatter/tests.rs
@@ -126,7 +126,7 @@ fn rustfmt_spec_is_correct() {
}
/// The edition on the command line overrides the project's own `rustfmt.toml`, so a constant
-/// there is not a default — it is Luxe overruling the repository about its own formatting.
+/// there is not a default — it is luxe overruling the repository about its own formatting.
#[test]
fn rust_is_formatted_against_the_edition_the_project_declares() {
let project = tempfile::tempdir().expect("temp dir");
@@ -150,7 +150,7 @@ fn rust_is_formatted_against_the_edition_the_project_declares() {
assert_eq!(edition_of(&FormatterRegistry::for_project(project.path(), false)), "2018");
// rustfmt reads `rustfmt.toml` itself, so what it says has to win over the manifest —
- // otherwise the flag Luxe passes silently overrules it.
+ // otherwise the flag luxe passes silently overrules it.
std::fs::write(project.path().join("rustfmt.toml"), "edition = \"2015\"\n").expect("write");
assert_eq!(edition_of(&FormatterRegistry::for_project(project.path(), false)), "2015");
}
diff --git a/crates/luxe-core/src/fs_atomic.rs b/crates/luxe-core/src/fs_atomic.rs
index b10c0dd..84788cd 100644
--- a/crates/luxe-core/src/fs_atomic.rs
+++ b/crates/luxe-core/src/fs_atomic.rs
@@ -57,7 +57,7 @@ pub fn replace_with_mode(path: &Path, bytes: &[u8], mode: Option) -> io::Re
/// Replace `path` keeping whatever mode the file already has, or owner-only when it is new.
///
/// For a config file that may hold a secret but that the user owns: a mode they chose stays theirs,
-/// and a file Luxe creates starts tight.
+/// and a file luxe creates starts tight.
pub fn replace_preserving_mode(path: &Path, bytes: &[u8]) -> io::Result<()> {
replace_with_mode(path, bytes, existing_mode(path))
}
diff --git a/crates/luxe-core/src/github.rs b/crates/luxe-core/src/github.rs
index 90cb1b4..8ac19e9 100644
--- a/crates/luxe-core/src/github.rs
+++ b/crates/luxe-core/src/github.rs
@@ -1,4 +1,4 @@
-//! GitHub — the one seam between Luxe and GitHub.
+//! GitHub — the one seam between luxe and GitHub.
//!
//! Everything that reads or writes a pull request or an issue goes through [`GitHub`], so there is
//! one module that knows `gh` exists. [`GhCli`] is the implementation: it runs the user's own `gh`
@@ -41,7 +41,7 @@ use async_trait::async_trait;
use serde::{Deserialize, Serialize};
use std::path::Path;
-/// What Luxe needs from a code host: every role, as the engine holds it.
+/// What luxe needs from a code host: every role, as the engine holds it.
///
/// Deliberately small: each method is something a caller in the tree actually uses, so the traits
/// grow with the features rather than ahead of them. The methods live in role traits, so code that
@@ -310,7 +310,7 @@ impl Repository {
/// [`classify_failure`], so rewording an error here never changes behavior somewhere else.
#[derive(Debug, thiserror::Error)]
pub enum GitHubError {
- /// `gh` is not installed anywhere Luxe can find it.
+ /// `gh` is not installed anywhere luxe can find it.
#[error("the GitHub CLI (`gh`) is not installed")]
NotInstalled,
/// `gh` is installed but has no working sign-in for the host.
@@ -340,8 +340,8 @@ pub enum GitHubError {
#[source]
source: std::io::Error,
},
- /// `gh` answered with output Luxe could not read.
- #[error("gh answered with something Luxe could not read")]
+ /// `gh` answered with output luxe could not read.
+ #[error("gh answered with something luxe could not read")]
Unreadable(#[source] serde_json::Error),
/// GitHub refused for rate: this sign-in's hourly budget, or its secondary limit, is spent.
#[error("GitHub's rate limit for this account is spent")]
diff --git a/crates/luxe-core/src/github/gh/tests/branches.rs b/crates/luxe-core/src/github/gh/tests/branches.rs
index 6dbf6d6..c602270 100644
--- a/crates/luxe-core/src/github/gh/tests/branches.rs
+++ b/crates/luxe-core/src/github/gh/tests/branches.rs
@@ -43,7 +43,7 @@ fn a_search_lists_pull_requests_and_issues_with_where_they_are() {
let deleted =
SEARCH_ISSUES.replacen("\"author\":{\"login\":", "\"author\":null,\"x\":{\"login\":", 1);
assert_eq!(parse_search(&deleted).unwrap().items[0].author, "", "a deleted account");
- // A frame from a Luxe that sent no state reads as open, not as an error
+ // A frame from a luxe that sent no state reads as open, not as an error
let older = r#"{"pull_request":true,"number":1,"title":"T","url":"u","repository":"a/b","author":"","draft":false,"updated_at":""}"#;
let item: ListItem = serde_json::from_str(older).unwrap();
assert_eq!(item.state, PullRequestState::Open);
@@ -72,7 +72,7 @@ fn what_merged_below_a_stack_is_read_with_where_it_went() {
pub(super) const STATUS_UNSTACKED: &str =
include_str!("../../fixtures/branch_status_unstacked.json");
-/// A real chain of Luxe's own, read from the middle: seven below down to `master`, two above,
+/// A real chain of luxe's own, read from the middle: seven below down to `master`, two above,
/// and the evidence that checks wait for trunk — the bottom one has them, this one none.
#[test]
fn a_pull_request_in_the_middle_of_a_stack_knows_where_it_sits() {
diff --git a/crates/luxe-core/src/github/gh/tests/threads.rs b/crates/luxe-core/src/github/gh/tests/threads.rs
index aabd37a..e4ffa4c 100644
--- a/crates/luxe-core/src/github/gh/tests/threads.rs
+++ b/crates/luxe-core/src/github/gh/tests/threads.rs
@@ -135,7 +135,7 @@ fn the_conversation_comes_with_the_threads() {
assert!(before.conversation.is_empty() && !before.threads.is_empty());
}
-/// On a real pull request the reader answered the first comment from Luxe, so it and the answer
+/// On a real pull request the reader answered the first comment from luxe, so it and the answer
/// stay quiet, and the review's summary is what is still asked.
#[test]
fn an_answered_comment_on_a_real_pull_request_asks_nothing() {
diff --git a/crates/luxe-core/src/github/overview.rs b/crates/luxe-core/src/github/overview.rs
index f351ece..4152633 100644
--- a/crates/luxe-core/src/github/overview.rs
+++ b/crates/luxe-core/src/github/overview.rs
@@ -321,7 +321,7 @@ pub struct IssueOverview {
pub conversation_count: usize,
}
-/// A session of yours that took an issue up, which only Luxe knows about.
+/// A session of yours that took an issue up, which only luxe knows about.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct LocalSession {
/// The session's id, which resuming it names
diff --git a/crates/luxe-core/src/github/refresh.rs b/crates/luxe-core/src/github/refresh.rs
index b2597a7..00c43d5 100644
--- a/crates/luxe-core/src/github/refresh.rs
+++ b/crates/luxe-core/src/github/refresh.rs
@@ -1,7 +1,7 @@
//! Refresh — when to ask GitHub about the branch's pull request again.
//!
//! GitHub has no push channel a local program can subscribe to, so the status row is kept current
-//! two ways: right away after anything Luxe saw happen, and on an interval otherwise. This module
+//! two ways: right away after anything luxe saw happen, and on an interval otherwise. This module
//! is the interval, and only the arithmetic of it — the workspace runtime tells it what happened
//! and asks when to look next, so every rule here is testable without a clock or a network.
//!
diff --git a/crates/luxe-core/src/github/remote/tests.rs b/crates/luxe-core/src/github/remote/tests.rs
index 8614232..adda66a 100644
--- a/crates/luxe-core/src/github/remote/tests.rs
+++ b/crates/luxe-core/src/github/remote/tests.rs
@@ -117,7 +117,7 @@ fn remotes_are_read_from_git_remote_v() {
fn the_matching_remote_is_found_by_where_it_points() {
let remotes = vec![
("origin".to_string(), "git@github.com:me/luxe.git".to_string()),
- ("upstream".to_string(), "https://github.com/Someone/Luxe".to_string()),
+ ("upstream".to_string(), "https://github.com/Someone/luxe".to_string()),
];
assert_eq!(
matching_remote(&remotes, "github.com", "someone", "luxe"),
diff --git a/crates/luxe-core/src/guided_review/attention.rs b/crates/luxe-core/src/guided_review/attention.rs
index 6ab9dd2..9661d82 100644
--- a/crates/luxe-core/src/guided_review/attention.rs
+++ b/crates/luxe-core/src/guided_review/attention.rs
@@ -4,7 +4,7 @@
//! for everything it can be expected to disclose and the wrong one for the rest. That line decides
//! what belongs here: a file edited without ever being read, a credential-shaped addition, a
//! weakened gate. The agent cannot report the first — it does not know — and has every incentive to
-//! soften the other two. Luxe can see all three because it is the harness that produced the change.
+//! soften the other two. luxe can see all three because it is the harness that produced the change.
//!
//! Who wrote a change is deliberately not here: read off the transcript it is a guess, wrong often
//! enough to cost more than it tells anyone, since a formatter, a pasted file, or a path spelled
diff --git a/crates/luxe-core/src/guided_review/context.rs b/crates/luxe-core/src/guided_review/context.rs
index 978bc11..5ef7617 100644
--- a/crates/luxe-core/src/guided_review/context.rs
+++ b/crates/luxe-core/src/guided_review/context.rs
@@ -114,8 +114,8 @@ impl ReviewContextDraft {
///
/// `extra_planning` names additional project-relative planning files (from `review.context_files`);
/// each is optional and read only when it resolves inside the opened project. The defaults are
- /// deliberately Luxe's own: a file belonging to the harness Luxe is *developed* with is excluded
- /// from release archives, so in any shipped Luxe it would never exist.
+ /// deliberately luxe's own: a file belonging to the harness luxe is *developed* with is excluded
+ /// from release archives, so in any shipped luxe it would never exist.
pub fn from_project_local_sources(
title: impl Into,
project_root: &Path,
diff --git a/crates/luxe-core/src/guided_review/context/tests.rs b/crates/luxe-core/src/guided_review/context/tests.rs
index c6311cc..904cce3 100644
--- a/crates/luxe-core/src/guided_review/context/tests.rs
+++ b/crates/luxe-core/src/guided_review/context/tests.rs
@@ -53,8 +53,8 @@ fn review_targets_round_trip_without_literal_diff_payloads() {
}
/// Context comes from the project's own artifacts. It used to read `.pi/agent-loop/PLAN.md` —
-/// a file belonging to the harness Luxe is developed with, excluded from release archives —
-/// so a shipped Luxe always produced empty excerpts.
+/// a file belonging to the harness luxe is developed with, excluded from release archives —
+/// so a shipped luxe always produced empty excerpts.
#[test]
fn draft_collects_project_local_sources_in_stable_categories() {
let project = tempfile::tempdir().unwrap();
diff --git a/crates/luxe-core/src/guided_review/exclude.rs b/crates/luxe-core/src/guided_review/exclude.rs
index af1d6cb..a84e2a7 100644
--- a/crates/luxe-core/src/guided_review/exclude.rs
+++ b/crates/luxe-core/src/guided_review/exclude.rs
@@ -1,6 +1,6 @@
//! Review exclusions — what a review leaves out, and why that is safe to do.
//!
-//! Two things a repository carries that nobody reviews: Luxe's own `.luxe/` directory (sessions,
+//! Two things a repository carries that nobody reviews: luxe's own `.luxe/` directory (sessions,
//! logs, caches, screenshots, task and review memory — byproducts of the tool running, not changes
//! to the project), and whatever bookkeeping a given project keeps that its reviewers already
//! ignore. The first is excluded always and needs no configuration; the second is `review.exclude`.
@@ -11,7 +11,7 @@
use serde::{Deserialize, Serialize};
-/// Luxe's own project directory. Never reviewed, never configurable.
+/// luxe's own project directory. Never reviewed, never configurable.
pub const LUXE_PROJECT_DIR: &str = ".luxe";
/// What a review leaves out of the diff it presents.
@@ -37,13 +37,13 @@ impl ReviewExclusions {
is_luxe_artifact(path) || self.patterns.iter().any(|pattern| matches_glob(pattern, path))
}
- /// Whether anything beyond Luxe's own directory is configured.
+ /// Whether anything beyond luxe's own directory is configured.
pub fn is_empty(&self) -> bool {
self.patterns.is_empty()
}
}
-/// Whether a path is a byproduct of Luxe running rather than a change to the project.
+/// Whether a path is a byproduct of luxe running rather than a change to the project.
pub fn is_luxe_artifact(path: &str) -> bool {
path == LUXE_PROJECT_DIR
|| path.starts_with(&format!("{LUXE_PROJECT_DIR}/"))
diff --git a/crates/luxe-core/src/guided_review/feedback_message.rs b/crates/luxe-core/src/guided_review/feedback_message.rs
index ac81c3d..b376143 100644
--- a/crates/luxe-core/src/guided_review/feedback_message.rs
+++ b/crates/luxe-core/src/guided_review/feedback_message.rs
@@ -84,7 +84,7 @@ pub fn format_review_feedback_message(
pub(crate) const REPORT_INSTRUCTION: &str =
"End your reply with this block, exactly as shown, one \
line per comment in the order above. Set each status to fixed, not_fixed or \
- needs_clarification, and replace each note with one sentence on what you did. Luxe reads it to \
+ needs_clarification, and replace each note with one sentence on what you did. luxe reads it to \
update the review; the reviewer sees your prose, not this.";
/// A finding's words as the agent reads them: the reader's own as they are, and a GitHub
diff --git a/crates/luxe-core/src/guided_review/feedback_message/tests.rs b/crates/luxe-core/src/guided_review/feedback_message/tests.rs
index 4d6703f..299a75d 100644
--- a/crates/luxe-core/src/guided_review/feedback_message/tests.rs
+++ b/crates/luxe-core/src/guided_review/feedback_message/tests.rs
@@ -138,7 +138,7 @@ fn formatter_keeps_review_evidence_deterministic_and_structured() {
assert!(message.contains("Do not rewrite unrelated code."));
// the report a fix request asks for is the last thing it says, with the real IDs in it,
- // and filled in it is a report Luxe accepts
+ // and filled in it is a report luxe accepts
let template = crate::guided_review::remediation_report_template(std::slice::from_ref(
&finding.finding_id,
));
diff --git a/crates/luxe-core/src/guided_review/feedback_session/finding.rs b/crates/luxe-core/src/guided_review/feedback_session/finding.rs
index e487ac4..512d7d9 100644
--- a/crates/luxe-core/src/guided_review/feedback_session/finding.rs
+++ b/crates/luxe-core/src/guided_review/feedback_session/finding.rs
@@ -176,7 +176,7 @@ pub enum FindingMessageAuthor {
Reviewer,
/// A remediation report from the active agent.
Agent,
- /// Luxe-generated lifecycle evidence.
+ /// luxe-generated lifecycle evidence.
System,
}
@@ -186,7 +186,7 @@ impl FindingMessageAuthor {
match self {
Self::Reviewer => "Reviewer",
Self::Agent => "Agent",
- Self::System => "Luxe",
+ Self::System => "luxe",
}
}
}
diff --git a/crates/luxe-core/src/guided_review/feedback_session/submissions.rs b/crates/luxe-core/src/guided_review/feedback_session/submissions.rs
index abb710c..6121624 100644
--- a/crates/luxe-core/src/guided_review/feedback_session/submissions.rs
+++ b/crates/luxe-core/src/guided_review/feedback_session/submissions.rs
@@ -187,7 +187,7 @@ impl ReviewSession {
Ok(())
}
- /// Notes, in Luxe's own name, that a fix turn ended with no word from the agent on these
+ /// Notes, in luxe's own name, that a fix turn ended with no word from the agent on these
/// comments, and why.
///
/// Without it a comment the agent was asked to fix looks exactly as if it had been ignored. The
diff --git a/crates/luxe-core/src/guided_review/guide_plan.rs b/crates/luxe-core/src/guided_review/guide_plan.rs
index 1bbf414..f29f886 100644
--- a/crates/luxe-core/src/guided_review/guide_plan.rs
+++ b/crates/luxe-core/src/guided_review/guide_plan.rs
@@ -420,7 +420,7 @@ const BRIEF_INSTRUCTION: &str = "Also add a top-level \"brief\": {\"intent\", \"
Its reader has already seen the conversation, so don't repeat it. `intent` is one or two \
sentences on the problem being solved. `account` is up to five short claims about what the \
change does, each one something a reviewer can check in the diff. Don't report on your own \
- testing; Luxe states that from the record.\n";
+ testing; luxe states that from the record.\n";
/// The request describing one batch, or `None` when it cannot be serialized.
fn batch_prompt(
diff --git a/crates/luxe-core/src/guided_review/loader/tests/capture_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/capture_tests.rs
index 5b3f7b2..0b9a28d 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/capture_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/capture_tests.rs
@@ -10,7 +10,7 @@ async fn repository() -> (tempfile::TempDir, LocalGit) {
let project = tempfile::tempdir().unwrap();
let git = LocalGit::new(project.path());
git.run("initialize repository", &["init", "--quiet"]).await.unwrap();
- git.run("name the author", &["config", "user.name", "Luxe Tests"]).await.unwrap();
+ git.run("name the author", &["config", "user.name", "luxe Tests"]).await.unwrap();
git.run("mail the author", &["config", "user.email", "tests@luxe.invalid"]).await.unwrap();
std::fs::write(project.path().join("keep.txt"), "kept\n").unwrap();
commit_all(&git, "initial").await;
@@ -154,7 +154,7 @@ async fn a_commit_of_empty_and_binary_additions_loads_whole() {
assert!(!review.chapters.is_empty());
}
-/// Luxe's own rendering of an empty file that is gone parses as the deletion it describes.
+/// luxe's own rendering of an empty file that is gone parses as the deletion it describes.
#[test]
fn an_empty_whole_file_deletion_parses() {
let rendered = crate::guided_review::render_whole_file_deletion(Path::new("gone.txt"), "");
diff --git a/crates/luxe-core/src/guided_review/loader/tests/commit_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/commit_tests.rs
index 887df29..556373d 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/commit_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/commit_tests.rs
@@ -11,7 +11,7 @@ async fn project_with_changes() -> (tempfile::TempDir, LocalGit) {
// In the repository, not as `-c` flags on the setup commit: the commit under test is made by
// *product* code, which rightly never invents an author. A machine with no global identity — a
// CI runner, a fresh container — then fails at the assertion rather than at the fixture.
- git.run("configure author", &["config", "user.name", "Luxe Tests"]).await.unwrap();
+ git.run("configure author", &["config", "user.name", "luxe Tests"]).await.unwrap();
git.run("configure author", &["config", "user.email", "tests@luxe.local"]).await.unwrap();
std::fs::write(project.path().join("lib.rs"), "let a = 1;\n").unwrap();
std::fs::write(project.path().join("notes.md"), "old\n").unwrap();
diff --git a/crates/luxe-core/src/guided_review/loader/tests/dispatch_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/dispatch_tests.rs
index 001f648..7c03f4b 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/dispatch_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/dispatch_tests.rs
@@ -181,7 +181,7 @@ async fn base_target_resolves_a_local_merge_base_and_rejects_unsafe_input() {
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -240,7 +240,7 @@ async fn commit_target_records_first_parent_and_root_semantics() {
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -273,7 +273,7 @@ async fn commit_target_records_first_parent_and_root_semantics() {
"create follow-up commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -315,7 +315,7 @@ async fn merge_commit_target_uses_only_its_first_parent() {
"commit base",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -333,7 +333,7 @@ async fn merge_commit_target_uses_only_its_first_parent() {
"commit feature",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -351,7 +351,7 @@ async fn merge_commit_target_uses_only_its_first_parent() {
"commit main",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -369,7 +369,7 @@ async fn merge_commit_target_uses_only_its_first_parent() {
"merge feature",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"merge",
@@ -539,7 +539,7 @@ async fn commit(git: &LocalGit, message: &str) {
"create commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
diff --git a/crates/luxe-core/src/guided_review/loader/tests/pull_request_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/pull_request_tests.rs
index 2d1579a..6afbf42 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/pull_request_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/pull_request_tests.rs
@@ -11,7 +11,7 @@ fn git(dir: &Path, args: &[&str]) -> String {
}
fn identify(dir: &Path) {
- git(dir, &["config", "user.name", "Luxe Tests"]);
+ git(dir, &["config", "user.name", "luxe Tests"]);
git(dir, &["config", "user.email", "tests@luxe.local"]);
}
diff --git a/crates/luxe-core/src/guided_review/loader/tests/source_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/source_tests.rs
index 3015246..6c9e23e 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/source_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/source_tests.rs
@@ -41,7 +41,7 @@ async fn complete_head_worktree_source_has_one_tracked_stream_and_sorted_untrack
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -292,7 +292,7 @@ async fn provenance_marks_staged_and_unstaged_changes_without_a_patch() {
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -324,7 +324,7 @@ async fn index_delete_and_worktree_recreate_remains_a_final_state_noop() {
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -364,7 +364,7 @@ async fn tracked_capture_compares_the_head_to_the_final_worktree_state() {
"create initial commit",
&[
"-c",
- "user.name=Luxe Tests",
+ "user.name=luxe Tests",
"-c",
"user.email=tests@luxe.invalid",
"commit",
@@ -453,7 +453,7 @@ async fn unborn_worktree_resolves_to_the_empty_tree() {
);
}
-/// Luxe's own directory is a byproduct of the tool running, never the user's change.
+/// luxe's own directory is a byproduct of the tool running, never the user's change.
///
/// It arrived in reviews two ways: as an untracked path (`.luxe/.gitignore` used to un-ignore
/// itself, which is all Git needs to report the whole directory), and — for anyone who committed
@@ -488,7 +488,7 @@ async fn an_untracked_file_the_policy_cannot_hold_is_left_out_and_named() {
git.run("stage initial file", &["add", "tracked.txt"]).await.unwrap();
git.run(
"commit initial file",
- &["-c", "user.name=Luxe", "-c", "user.email=luxe@example.invalid", "commit", "-qm", "base"],
+ &["-c", "user.name=luxe", "-c", "user.email=luxe@example.invalid", "commit", "-qm", "base"],
)
.await
.unwrap();
diff --git a/crates/luxe-core/src/guided_review/loader/tests/turn_tests.rs b/crates/luxe-core/src/guided_review/loader/tests/turn_tests.rs
index dd1ff15..26e7f52 100644
--- a/crates/luxe-core/src/guided_review/loader/tests/turn_tests.rs
+++ b/crates/luxe-core/src/guided_review/loader/tests/turn_tests.rs
@@ -39,7 +39,7 @@ async fn last_turn_preserves_preexisting_dirty_and_untracked_content() {
git.run("configure test identity", &["config", "user.email", "tests@example.invalid"])
.await
.unwrap();
- git.run("configure test name", &["config", "user.name", "Luxe Tests"]).await.unwrap();
+ git.run("configure test name", &["config", "user.name", "luxe Tests"]).await.unwrap();
std::fs::write(project.path().join("tracked.txt"), "committed\n").unwrap();
git.run("stage fixture", &["add", "tracked.txt"]).await.unwrap();
git.run("commit fixture", &["commit", "--quiet", "-m", "fixture"]).await.unwrap();
diff --git a/crates/luxe-core/src/guided_review/model.rs b/crates/luxe-core/src/guided_review/model.rs
index 6e2326e..fc8378e 100644
--- a/crates/luxe-core/src/guided_review/model.rs
+++ b/crates/luxe-core/src/guided_review/model.rs
@@ -58,7 +58,7 @@ pub enum ReviewBase {
/// First parent, absent for a root commit.
parent_oid: Option,
},
- /// Complete pre-turn baseline retained by Luxe.
+ /// Complete pre-turn baseline retained by luxe.
TurnBaseline {
/// Immutable baseline revision identifier.
revision_id: ReviewRevisionId,
@@ -201,7 +201,7 @@ pub struct ReviewStats {
pub binary_files: u64,
/// Number of explicit non-text file items.
pub non_text_files: u64,
- /// Files left out of the review by `review.exclude` or as Luxe's own artifacts.
+ /// Files left out of the review by `review.exclude` or as luxe's own artifacts.
///
/// Counted, never silent: a review that shows less than the diff has to say so.
#[serde(default)]
diff --git a/crates/luxe-core/src/guided_review/posting.rs b/crates/luxe-core/src/guided_review/posting.rs
index a15dc43..4ea2ff7 100644
--- a/crates/luxe-core/src/guided_review/posting.rs
+++ b/crates/luxe-core/src/guided_review/posting.rs
@@ -2,7 +2,7 @@
//!
//! The reviewer's comments become a single review with a verdict, the way they would submit one on
//! the site. Where each comment lands is decided here, against GitHub's own patches rather than
-//! Luxe's diff: the two usually agree, and a comment that assumed so and was wrong would be refused
+//! luxe's diff: the two usually agree, and a comment that assumed so and was wrong would be refused
//! (`Line could not be resolved`) and take the whole review with it.
//!
//! Nothing is added to the reviewer's words. A comment that cannot sit on a line goes in the
diff --git a/crates/luxe-core/src/guided_review/posting/tests.rs b/crates/luxe-core/src/guided_review/posting/tests.rs
index 7fe1f57..26ee9a1 100644
--- a/crates/luxe-core/src/guided_review/posting/tests.rs
+++ b/crates/luxe-core/src/guided_review/posting/tests.rs
@@ -48,7 +48,7 @@ fn review() -> GuidedReview {
}
}
-/// GitHub's patch agrees with Luxe's diff unless a test says otherwise.
+/// GitHub's patch agrees with luxe's diff unless a test says otherwise.
fn patches_matching(review: &GuidedReview) -> Vec {
review
.files
diff --git a/crates/luxe-core/src/guided_review/provenance.rs b/crates/luxe-core/src/guided_review/provenance.rs
index 99becd8..8941566 100644
--- a/crates/luxe-core/src/guided_review/provenance.rs
+++ b/crates/luxe-core/src/guided_review/provenance.rs
@@ -2,7 +2,7 @@
//!
//! A diff shows the result; the process is where the review-critical signals live. A skipped test,
//! a file edited without ever being read, a weakened gate, work nobody asked for — none of those
-//! appear in the patch. Luxe is the harness, so it can read them off its own transcript instead of
+//! appear in the patch. luxe is the harness, so it can read them off its own transcript instead of
//! inferring intent from filenames.
//!
//! This is bounded, redacted, UI-neutral data. It grounds the chaptering prompt and renders as the
@@ -53,7 +53,7 @@ impl CommandRun {
}
}
-/// Whether a command line reads as a test run across the ecosystems Luxe drives.
+/// Whether a command line reads as a test run across the ecosystems luxe drives.
fn looks_like_test_command(command: &str) -> bool {
let lowered = command.to_ascii_lowercase();
[
@@ -118,7 +118,7 @@ impl TurnProvenance {
match message.role {
Role::User => {
if let Some(text) = message_text(message) {
- // Feedback submissions are Luxe talking to itself, not a user request.
+ // Feedback submissions are luxe talking to itself, not a user request.
if !text.starts_with("Guided review feedback") {
provenance.requests.push(bounded(&text, MAX_ACCOUNT_CHARS));
}
diff --git a/crates/luxe-core/src/guided_review/provenance/tests.rs b/crates/luxe-core/src/guided_review/provenance/tests.rs
index 6503728..8d49b62 100644
--- a/crates/luxe-core/src/guided_review/provenance/tests.rs
+++ b/crates/luxe-core/src/guided_review/provenance/tests.rs
@@ -204,7 +204,7 @@ fn a_path_outside_the_project_is_not_recorded() {
assert!(provenance.written_paths.is_empty(), "nothing outside the project is attributed");
}
-/// The `test` tool is the one Luxe tells models to prefer, and its command is auto-detected,
+/// The `test` tool is the one luxe tells models to prefer, and its command is auto-detected,
/// so an argument-less call still ran the suite.
#[test]
fn the_test_tool_counts_as_a_command_that_verifies() {
diff --git a/crates/luxe-core/src/guided_review/pull_request_draft.rs b/crates/luxe-core/src/guided_review/pull_request_draft.rs
index 447d3b5..37134cf 100644
--- a/crates/luxe-core/src/guided_review/pull_request_draft.rs
+++ b/crates/luxe-core/src/guided_review/pull_request_draft.rs
@@ -3,10 +3,10 @@
//! A branch's Guided Review already says what changed and why, chapter by chapter, in the order it
//! makes sense. That is the description a pull request usually lacks, so it becomes the draft: the
//! review's intent, each chapter with its reasons, and `Closes #n` when the session took up an
-//! issue. Anyone reading it on GitHub gets the story, with or without Luxe.
+//! issue. Anyone reading it on GitHub gets the story, with or without luxe.
//!
//! It is a draft. The reader edits it in a preview before anything is sent, and nothing is added
-//! that they did not see: no signature, no marker, nothing about Luxe.
+//! that they did not see: no signature, no marker, nothing about luxe.
use super::{ChapterKind, GuidedReview};
diff --git a/crates/luxe-core/src/guided_review/remediation.rs b/crates/luxe-core/src/guided_review/remediation.rs
index f34f360..4d3489a 100644
--- a/crates/luxe-core/src/guided_review/remediation.rs
+++ b/crates/luxe-core/src/guided_review/remediation.rs
@@ -84,13 +84,13 @@ pub enum RemediationReportError {
/// The report describes a different immutable source revision.
#[error("remediation report has a different source revision")]
WrongRevision,
- /// Not one entry names a comment that was sent with a status Luxe reads.
+ /// Not one entry names a comment that was sent with a status luxe reads.
#[error("remediation report names no submitted finding with a known status")]
NothingUsable,
}
impl RemediationReportError {
- /// What went wrong, as the tail of a sentence: told to the agent when Luxe asks again, and to
+ /// What went wrong, as the tail of a sentence: told to the agent when luxe asks again, and to
/// the reviewer on each comment the report did not cover.
pub fn reason(&self) -> &'static str {
match self {
@@ -232,7 +232,7 @@ fn first_json_value(text: &str) -> Result<(Value, usize), serde_json::Error> {
///
/// A review or revision the report names must be the turn's, prefix or no prefix, or the whole
/// report is refused: it is about something else. One it leaves out is the turn's. Inside it, an entry for a comment that was not sent, a second entry for
-/// the same comment, or a status Luxe cannot read is dropped and the rest are kept. A report left
+/// the same comment, or a status luxe cannot read is dropped and the rest are kept. A report left
/// with nothing is [`RemediationReportError::NothingUsable`].
fn read_report(
value: &Value,
@@ -260,7 +260,7 @@ fn read_report(
findings.push(result);
}
Some(_) => tracing::debug!("dropped a second report entry for one comment"),
- None => tracing::debug!(%entry, "dropped a report entry Luxe cannot use"),
+ None => tracing::debug!(%entry, "dropped a report entry luxe cannot use"),
}
}
if findings.is_empty() {
@@ -273,7 +273,7 @@ fn read_report(
})
}
-/// One entry, when it names a comment that was sent and has a status Luxe reads.
+/// One entry, when it names a comment that was sent and has a status luxe reads.
fn read_result(entry: &Value, known: &[FindingId]) -> Option {
let named = entry.get("finding_id")?.as_str()?.trim();
// the known ID, never one built out of what the model wrote
diff --git a/crates/luxe-core/src/guided_review/remediation/tests.rs b/crates/luxe-core/src/guided_review/remediation/tests.rs
index 4aa959a..9d56b46 100644
--- a/crates/luxe-core/src/guided_review/remediation/tests.rs
+++ b/crates/luxe-core/src/guided_review/remediation/tests.rs
@@ -26,7 +26,7 @@ fn extract(text: &str) -> Result, RemediationReportErr
fn statuses(text: &str) -> Vec<(String, RemediationStatus)> {
extract(text)
- .expect("a report Luxe can use")
+ .expect("a report luxe can use")
.expect("a report")
.findings
.into_iter()
diff --git a/crates/luxe-core/src/guided_review/seen_store.rs b/crates/luxe-core/src/guided_review/seen_store.rs
index 1ca33e9..0e3f070 100644
--- a/crates/luxe-core/src/guided_review/seen_store.rs
+++ b/crates/luxe-core/src/guided_review/seen_store.rs
@@ -207,7 +207,7 @@ mod tests {
number,
repository: repository.map(str::to_owned),
};
- store.record(&pr(7, Some("Upstream/Luxe")), seen(&["a"])).unwrap();
+ store.record(&pr(7, Some("Upstream/luxe")), seen(&["a"])).unwrap();
store.record(&pr(7, None), seen(&["b"])).unwrap();
assert_eq!(store.load(&pr(7, Some("upstream/luxe"))).unwrap().item_ids, vec!["a"]);
diff --git a/crates/luxe-core/src/guided_review/source_store.rs b/crates/luxe-core/src/guided_review/source_store.rs
index c43a01f..14a6167 100644
--- a/crates/luxe-core/src/guided_review/source_store.rs
+++ b/crates/luxe-core/src/guided_review/source_store.rs
@@ -90,7 +90,7 @@ pub enum SourceStoreError {
/// Underlying JSON error.
error: serde_json::Error,
},
- /// The artifact uses a schema this Luxe version does not understand.
+ /// The artifact uses a schema this luxe version does not understand.
#[error("review source artifact uses unsupported schema {schema_version}")]
UnsupportedSchema {
/// Schema version found in the manifest.
diff --git a/crates/luxe-core/src/hook.rs b/crates/luxe-core/src/hook.rs
index db3a928..d5b6d9f 100644
--- a/crates/luxe-core/src/hook.rs
+++ b/crates/luxe-core/src/hook.rs
@@ -87,7 +87,7 @@ impl HookPoint {
Self::PermissionResolved,
];
- /// The one spelling Luxe writes and documents, matching how the variant serializes.
+ /// The one spelling luxe writes and documents, matching how the variant serializes.
///
/// Exhaustive on purpose: a new variant does not compile until it has a name here.
pub fn canonical_name(&self) -> &'static str {
diff --git a/crates/luxe-core/src/hooks/herdr.rs b/crates/luxe-core/src/hooks/herdr.rs
index ee9a9e7..a493350 100644
--- a/crates/luxe-core/src/hooks/herdr.rs
+++ b/crates/luxe-core/src/hooks/herdr.rs
@@ -5,9 +5,9 @@
//! at a glance, which agent is running, which needs a decision, and which is ready to review.
//!
//! Herdr recognizes most agents by *process detection* + screen scraping, which requires a Herdr
-//! binary update per new agent. Instead Luxe acts as a **custom socket integration** (Herdr's
-//! supported extension path): when Luxe runs inside a Herdr pane (`HERDR_ENV=1`), this hook reports
-//! Luxe's own lifecycle to Herdr via the `herdr pane report-agent` CLI. That makes Luxe a
+//! binary update per new agent. Instead luxe acts as a **custom socket integration** (Herdr's
+//! supported extension path): when luxe runs inside a Herdr pane (`HERDR_ENV=1`), this hook reports
+//! luxe's own lifecycle to Herdr via the `herdr pane report-agent` CLI. That makes luxe a
//! first-class, correctly-tracked agent in the sidebar without Herdr needing to know the `luxe`
//! process at all.
//!
@@ -136,7 +136,7 @@ impl HerdrReporter for CliHerdrReporter {
}
}
-/// Reports Luxe's lifecycle to Herdr. Registered only when running inside a Herdr pane.
+/// Reports luxe's lifecycle to Herdr. Registered only when running inside a Herdr pane.
pub struct HerdrHook {
reporter: Box,
/// Last state sent, to suppress redundant reports (a turn fires `AgentStart` many times).
diff --git a/crates/luxe-core/src/hooks/plan_reminder.rs b/crates/luxe-core/src/hooks/plan_reminder.rs
index aae251b..7f5e831 100644
--- a/crates/luxe-core/src/hooks/plan_reminder.rs
+++ b/crates/luxe-core/src/hooks/plan_reminder.rs
@@ -271,7 +271,7 @@ mod tests {
assert_eq!(run(&hook, "read", 1).await.len(), 1);
}
- /// Loading a deferred tool's schema is Luxe's own request-budget bookkeeping, not a step in the
+ /// Loading a deferred tool's schema is luxe's own request-budget bookkeeping, not a step in the
/// task — so it cannot be the work a plan has fallen behind.
#[tokio::test]
async fn bookkeeping_calls_are_not_work_the_plan_fell_behind() {
diff --git a/crates/luxe-core/src/hooks/self_continuation.rs b/crates/luxe-core/src/hooks/self_continuation.rs
index 0691159..2b98f14 100644
--- a/crates/luxe-core/src/hooks/self_continuation.rs
+++ b/crates/luxe-core/src/hooks/self_continuation.rs
@@ -273,11 +273,23 @@ impl Hook for SelfContinuationHook {
info!(iteration, "continuing the loop");
debug!(%steer, "self-continuation steer");
// The one marker that makes a run readable from outside: where it is, each time it moves.
- cx.notify(format!("iteration {iteration} · phase {}", state.current_phase));
+ cx.notify(iteration_notice(iteration, state.current_phase));
HookOutcome::ContinueLoop { message: steer }
}
}
+/// The notice that opens each iteration: `iteration 4 · phase 1`.
+pub fn iteration_notice(iteration: u64, phase: u32) -> String {
+ format!("iteration {iteration} · phase {phase}")
+}
+
+/// The iteration and phase an [`iteration_notice`] names, or `None` for any other notice. Beside
+/// the writer so a front-end drawing the run's structure never guesses at the wording.
+pub fn parse_iteration_notice(notice: &str) -> Option<(u64, u32)> {
+ let (iteration, phase) = notice.strip_prefix("iteration ")?.split_once(" · phase ")?;
+ Some((iteration.parse().ok()?, phase.parse().ok()?))
+}
+
#[cfg(test)]
mod tests;
diff --git a/crates/luxe-core/src/hooks/self_continuation/tests.rs b/crates/luxe-core/src/hooks/self_continuation/tests.rs
index adfc8bd..ae7677a 100644
--- a/crates/luxe-core/src/hooks/self_continuation/tests.rs
+++ b/crates/luxe-core/src/hooks/self_continuation/tests.rs
@@ -209,3 +209,24 @@ async fn test_self_continuation_hook_stopped_sentinel() {
// STOP sentinel exists -> Continue.
assert!(matches!(outcome, HookOutcome::Continue));
}
+
+/// The notice is pinned as a literal, since the loop view reads it, and the parser reads it back.
+#[test]
+fn the_iteration_notice_reads_back() {
+ assert_eq!(iteration_notice(4, 1), "iteration 4 · phase 1");
+ assert_eq!(parse_iteration_notice("iteration 4 · phase 1"), Some((4, 1)));
+}
+
+/// Every other notice, a near miss included, is not an iteration.
+#[test]
+fn other_notices_are_not_iterations() {
+ for notice in [
+ "phase 0 accepted — now on phase 1",
+ "iteration 4",
+ "iteration x · phase 1",
+ "iteration 4 · phase one",
+ "",
+ ] {
+ assert_eq!(parse_iteration_notice(notice), None, "{notice}");
+ }
+}
diff --git a/crates/luxe-core/src/instructions.rs b/crates/luxe-core/src/instructions.rs
index 8ed3eb7..217539c 100644
--- a/crates/luxe-core/src/instructions.rs
+++ b/crates/luxe-core/src/instructions.rs
@@ -71,7 +71,7 @@ impl Sanitized {
}
if self.withheld {
return Some(format!(
- "{name}: left out of the prompt — it repeats link definitions more often than Luxe \
+ "{name}: left out of the prompt — it repeats link definitions more often than luxe \
can check for hidden text"
));
}
@@ -132,7 +132,7 @@ const PLACEHOLDER: char = '\u{7f}';
/// The whole text of a withheld file, so every caller passes the reason on rather than an empty
/// string.
-const WITHHELD: &str = "[withheld: this file repeats link definitions more often than Luxe can \
+const WITHHELD: &str = "[withheld: this file repeats link definitions more often than luxe can \
check for hidden text]";
/// Characters that put text on the page without putting it in front of anyone.
@@ -175,7 +175,7 @@ pub enum ContextScope {
Global,
/// The top of the checkout, or the project directory when there is none
Project,
- /// A folder below that, on the way down to where Luxe was launched
+ /// A folder below that, on the way down to where luxe was launched
Folder,
}
@@ -226,7 +226,7 @@ pub fn context_files(cwd: &std::path::Path) -> Vec {
/// Marks each file that repeats an earlier one, by identity first and then by text.
///
-/// Byte-identical text is the only overlap Luxe settles on its own. Two files that say *mostly*
+/// Byte-identical text is the only overlap luxe settles on its own. Two files that say *mostly*
/// the same thing are two files someone wrote, and choosing between their sentences is the
/// author's job.
fn dedupe(found: impl Iterator- ) -> Vec
{
@@ -263,15 +263,15 @@ pub fn project_context_file(cwd: &std::path::Path) -> std::path::PathBuf {
top.unwrap_or_else(|| cwd.to_path_buf()).join(INSTRUCTION_FILES[0])
}
-/// The global instruction file Luxe creates when asked: `~/.config/luxe/agents.md`.
+/// The global instruction file luxe creates when asked: `~/.config/luxe/agents.md`.
pub fn global_context_file() -> std::path::PathBuf {
crate::user_dirs::luxe_config_dir().join("agents.md")
}
-/// Every global instruction file Luxe reads, broadest first, whether or not it exists.
+/// Every global instruction file luxe reads, broadest first, whether or not it exists.
///
/// `~/.agents/AGENTS.md` is the file a growing number of people keep for *every* agent they run,
-/// and Luxe's own sits after it so the Luxe-specific word lands later. Somebody who symlinked one to
+/// and luxe's own sits after it so the luxe-specific word lands later. Somebody who symlinked one to
/// the other gets it once.
pub fn global_context_files() -> Vec {
let shared = crate::user_dirs::home_dir().map(|home| home.join(".agents").join("AGENTS.md"));
diff --git a/crates/luxe-core/src/instructions/nested.rs b/crates/luxe-core/src/instructions/nested.rs
index c565839..9d4b2a6 100644
--- a/crates/luxe-core/src/instructions/nested.rs
+++ b/crates/luxe-core/src/instructions/nested.rs
@@ -1,7 +1,7 @@
//! Nested instructions — the `AGENTS.md` files off the launch chain, read when work reaches them.
//!
-//! The system prompt carries the files from the top of the checkout down to where Luxe started,
-//! and it's written once per session. A file in a folder off that chain (`web/AGENTS.md` when Luxe
+//! The system prompt carries the files from the top of the checkout down to where luxe started,
+//! and it's written once per session. A file in a folder off that chain (`web/AGENTS.md` when luxe
//! started at the root) can't join it without rewriting the prompt every later request is cached
//! against. So it rides on the result of the first tool call that reaches its folder: an append,
//! like any other result. Whether the model already has one is read off the conversation by the
diff --git a/crates/luxe-core/src/instructions/nested/tests.rs b/crates/luxe-core/src/instructions/nested/tests.rs
index 1bc6796..a3caf58 100644
--- a/crates/luxe-core/src/instructions/nested/tests.rs
+++ b/crates/luxe-core/src/instructions/nested/tests.rs
@@ -12,7 +12,7 @@ fn checkout(files: &[&str]) -> tempfile::TempDir {
dir
}
-/// What [`nested_files`] names for `target`, from Luxe launched at `launch`, both under `root`.
+/// What [`nested_files`] names for `target`, from luxe launched at `launch`, both under `root`.
fn shown(root: &Path, launch: &str, target: &str) -> Vec {
let files = nested_files(&root.join(launch), &root.join(target));
files.into_iter().map(|file| file.shown).collect()
diff --git a/crates/luxe-core/src/instructions/tests.rs b/crates/luxe-core/src/instructions/tests.rs
index 44646cb..cf03c59 100644
--- a/crates/luxe-core/src/instructions/tests.rs
+++ b/crates/luxe-core/src/instructions/tests.rs
@@ -649,7 +649,7 @@ fn identical_text_is_carried_once_and_different_text_is_not_touched() {
assert!(found[0].is_loaded());
assert!(!found[1].is_loaded());
assert_eq!(found[1].duplicate_of, Some(root.join("AGENTS.md")), "a byte-identical copy");
- assert!(found[2].is_loaded(), "overlap is the author's to settle, not Luxe's");
+ assert!(found[2].is_loaded(), "overlap is the author's to settle, not luxe's");
}
/// Two empty files are not "the same instructions" in any sense worth reporting.
@@ -682,7 +682,7 @@ fn a_directory_prefers_agents_md_but_accepts_claude_md() {
fn the_global_files_put_the_shared_one_first() {
let files = global_context_files();
assert!(global_context_file().ends_with(std::path::Path::new("luxe").join("agents.md")));
- assert_eq!(files.last(), Some(&global_context_file()), "Luxe's own comes last");
+ assert_eq!(files.last(), Some(&global_context_file()), "luxe's own comes last");
if let Some(home) = crate::user_dirs::home_dir() {
assert_eq!(files.first(), Some(&home.join(".agents").join("AGENTS.md")));
}
diff --git a/crates/luxe-core/src/json_file.rs b/crates/luxe-core/src/json_file.rs
index 7080587..9f7d8cd 100644
--- a/crates/luxe-core/src/json_file.rs
+++ b/crates/luxe-core/src/json_file.rs
@@ -2,10 +2,10 @@
//!
//! The provider credential store, the config and permissions writers and the MCP token store all
//! edit a file a human owns, which sets the rules: never lose a key we don't understand, never
-//! leave a half-written file behind, and never race another Luxe process (a token refresh in one
+//! leave a half-written file behind, and never race another luxe process (a token refresh in one
//! and a `/login` in another).
//!
-//! The lock is a directory. `create_dir` is atomic everywhere Luxe runs and needs no dependency; a
+//! The lock is a directory. `create_dir` is atomic everywhere luxe runs and needs no dependency; a
//! lock older than `LOCK_STALE` is assumed to belong to a killed process and is broken, because
//! refusing to ever write again is the worse failure.
@@ -88,7 +88,7 @@ impl JsonFile {
/// Parse as a JSON object, treating an absent, empty, or malformed file as "nothing yet".
///
- /// Reads are forgiving on purpose: a syntax error in a config file must not stop Luxe from
+ /// Reads are forgiving on purpose: a syntax error in a config file must not stop luxe from
/// starting, only from writing.
pub fn read(&self) -> Option> {
let text = std::fs::read_to_string(&self.path).ok()?;
@@ -124,7 +124,7 @@ impl JsonFile {
Ok(outcome)
}
- /// [`Self::modify`] for a file Luxe writes for itself, where one that does not parse is moved
+ /// [`Self::modify`] for a file luxe writes for itself, where one that does not parse is moved
/// aside rather than refused: to `.bad`, or `.bad-` when that is taken, so an
/// earlier one survives. The move happens under the lock, so two writers meeting the same broken
/// file cannot set each other's work aside. Returns where the broken file went.
diff --git a/crates/luxe-core/src/local_endpoint.rs b/crates/luxe-core/src/local_endpoint.rs
index bbcb52b..64deb48 100644
--- a/crates/luxe-core/src/local_endpoint.rs
+++ b/crates/luxe-core/src/local_endpoint.rs
@@ -1,6 +1,6 @@
-//! local_endpoint — the way into a running Luxe from the same machine, and the list of them.
+//! local_endpoint — the way into a running luxe from the same machine, and the list of them.
//!
-//! Each interactive Luxe serves its protocol here, so another front-end (the `luxe remote` bridge,
+//! Each interactive luxe serves its protocol here, so another front-end (the `luxe remote` bridge,
//! `luxe join`, a script) can join the session it is running. It is local IPC, not network I/O: a
//! Unix socket in a folder only this user can open, or on Windows a named pipe only this user can
//! open and no other machine can reach. This module is the one owner of those platform details.
@@ -28,14 +28,14 @@ pub trait LocalIo: AsyncRead + AsyncWrite + Unpin + Send {}
impl LocalIo for T {}
-/// A connection to or from a running Luxe.
+/// A connection to or from a running luxe.
pub type LocalStream = Box;
/// A client that connected, with its process id where the platform reports one.
pub struct Joined {
/// The connection
pub stream: LocalStream,
- /// The client's process id, for the check that it is not one of this Luxe's own children
+ /// The client's process id, for the check that it is not one of this luxe's own children
pub peer: Option,
}
@@ -61,7 +61,7 @@ impl EndpointError {
}
}
-/// This Luxe's endpoint, open for clients until it is dropped.
+/// This luxe's endpoint, open for clients until it is dropped.
pub struct EndpointListener(platform::Listener);
impl EndpointListener {
diff --git a/crates/luxe-core/src/local_endpoint/ancestry.rs b/crates/luxe-core/src/local_endpoint/ancestry.rs
index e622d22..e57ef69 100644
--- a/crates/luxe-core/src/local_endpoint/ancestry.rs
+++ b/crates/luxe-core/src/local_endpoint/ancestry.rs
@@ -1,6 +1,6 @@
//! Ancestry — whether a process was started, at any depth, by another.
//!
-//! The tripwire that keeps a Luxe's own children (the agent's `bash`, a hook) from joining its
+//! The tripwire that keeps a luxe's own children (the agent's `bash`, a hook) from joining its
//! session and answering their own prompts. A tripwire, not a wall: a process that detaches itself
//! leaves the chain, and the documentation says so. Blocking: run it off the async runtime.
diff --git a/crates/luxe-core/src/local_endpoint/live.rs b/crates/luxe-core/src/local_endpoint/live.rs
index 2959cec..111bd15 100644
--- a/crates/luxe-core/src/local_endpoint/live.rs
+++ b/crates/luxe-core/src/local_endpoint/live.rs
@@ -1,17 +1,17 @@
//! Live sessions — which Luxes on this machine can be joined, and where.
//!
-//! Each running Luxe with an endpoint leaves a small record in the runtime folder, named for its
-//! process, and takes it away when it stops. A record left by a Luxe that died is found out by the
+//! Each running luxe with an endpoint leaves a small record in the runtime folder, named for its
+//! process, and takes it away when it stops. A record left by a luxe that died is found out by the
//! one test that matters: its endpoint no longer answers.
use serde::{Deserialize, Serialize};
use std::io;
use std::path::{Path, PathBuf};
-/// One running Luxe that can be joined.
+/// One running luxe that can be joined.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct LiveSession {
- /// The Luxe's process id
+ /// The luxe's process id
pub pid: u32,
/// Session it is running
pub session_id: String,
@@ -23,7 +23,7 @@ pub struct LiveSession {
pub luxe_version: String,
}
-/// This Luxe's record in the runtime folder, removed when dropped.
+/// This luxe's record in the runtime folder, removed when dropped.
pub struct Publication {
path: PathBuf,
}
@@ -39,7 +39,7 @@ impl LiveSession {
}
impl Publication {
- /// Rewrites the record, as when the Luxe moves to another session.
+ /// Rewrites the record, as when the luxe moves to another session.
pub fn update(&self, session: &LiveSession) -> io::Result<()> {
let bytes = serde_json::to_vec_pretty(session).map_err(io::Error::other)?;
crate::fs_atomic::replace_private(&self.path, &bytes)
@@ -64,7 +64,7 @@ pub fn live_sessions(dir: &Path) -> Vec {
found
}
-/// [`live_sessions`] whose Luxe is still running: one that crashed leaves its record behind.
+/// [`live_sessions`] whose luxe is still running: one that crashed leaves its record behind.
///
/// Checked by process id rather than by connecting, since a connection is a client joining, and the
/// terminal running the session says so each time.
diff --git a/crates/luxe-core/src/local_endpoint/tests.rs b/crates/luxe-core/src/local_endpoint/tests.rs
index 0376553..af04d5a 100644
--- a/crates/luxe-core/src/local_endpoint/tests.rs
+++ b/crates/luxe-core/src/local_endpoint/tests.rs
@@ -35,7 +35,7 @@ async fn the_socket_is_private_and_goes_with_the_listener() {
assert_eq!(mode(&run), 0o700, "the folder");
assert_eq!(mode(&socket), 0o600, "the socket");
drop(listener);
- assert!(!socket.exists(), "a stopped Luxe leaves no socket behind");
+ assert!(!socket.exists(), "a stopped luxe leaves no socket behind");
assert!(connect(&socket.to_string_lossy()).await.is_err());
}
diff --git a/crates/luxe-core/src/local_endpoint/windows.rs b/crates/luxe-core/src/local_endpoint/windows.rs
index 286744e..66ff94c 100644
--- a/crates/luxe-core/src/local_endpoint/windows.rs
+++ b/crates/luxe-core/src/local_endpoint/windows.rs
@@ -4,7 +4,7 @@
//! the anonymous account, and reading is the whole transcript; and a pipe can be opened across the
//! network while the Server service runs. So every instance is created with a descriptor granting
//! this user alone, remote clients are refused, and the first instance must be new, so nothing can
-//! claim the name before Luxe does.
+//! claim the name before luxe does.
use super::{EndpointError, Joined, LocalStream};
use std::path::Path;
diff --git a/crates/luxe-core/src/loop_controller.rs b/crates/luxe-core/src/loop_controller.rs
index 19f23a0..1965cdf 100644
--- a/crates/luxe-core/src/loop_controller.rs
+++ b/crates/luxe-core/src/loop_controller.rs
@@ -349,7 +349,7 @@ impl LoopStatus {
///
/// The flag alone cannot answer it: a killed run leaves `running: true` behind and has no way
/// to correct it. An unrecorded owner reads as running, because that is what the file says and
- /// there is nothing to contradict it — a state written by an older Luxe, or a run marked
+ /// there is nothing to contradict it — a state written by an older luxe, or a run marked
/// resumable and not yet started.
pub fn is_live(&self) -> bool {
self.running && !matches!(self.liveness, crate::loop_state::LoopLiveness::Gone { .. })
diff --git a/crates/luxe-core/src/loop_state.rs b/crates/luxe-core/src/loop_state.rs
index df6f36d..b76d732 100644
--- a/crates/luxe-core/src/loop_state.rs
+++ b/crates/luxe-core/src/loop_state.rs
@@ -116,7 +116,7 @@ pub enum LoopLiveness {
/// The machine that wrote the record
host: String,
},
- /// Nothing was recorded: a state file written before Luxe kept this, or a run marked
+ /// Nothing was recorded: a state file written before luxe kept this, or a run marked
/// resumable by `luxe loop resume`, which flags the state without driving anything.
Unrecorded,
}
@@ -328,7 +328,7 @@ mod owner_tests {
assert_eq!(state.owned_by_another_process(), None);
}
- /// A state file written before Luxe recorded an owner still loads, and claims nothing.
+ /// A state file written before luxe recorded an owner still loads, and claims nothing.
#[test]
fn a_state_without_an_owner_makes_no_claim() {
let older = r#"{"running":true,"iteration":3,"current_phase":1,"max_turns_per_run":50,
diff --git a/crates/luxe-core/src/mcp_client.rs b/crates/luxe-core/src/mcp_client.rs
index d3a417f..732d434 100644
--- a/crates/luxe-core/src/mcp_client.rs
+++ b/crates/luxe-core/src/mcp_client.rs
@@ -32,7 +32,7 @@ pub enum McpTransport {
/// Working directory for the process.
cwd: Option,
/// Extra environment variables for the child process. Values support `${VAR}` and
- /// `$env:VAR` interpolation against Luxe's own environment.
+ /// `$env:VAR` interpolation against luxe's own environment.
#[serde(default)]
env: std::collections::HashMap,
},
@@ -114,7 +114,7 @@ pub enum McpAuth {
///
/// Spelled `oauth`, as every example in the docs writes it. `snake_case` alone made it
/// `o_auth`, so a hand-written `"kind": "oauth"` failed to parse and took the whole file with it;
- /// `o_auth` is still read, since that is what Luxe itself would have written.
+ /// `o_auth` is still read, since that is what luxe itself would have written.
#[serde(rename = "oauth", alias = "o_auth")]
OAuth(OAuthConfig),
}
@@ -276,7 +276,7 @@ pub struct McpToolResult {
pub error: Option,
}
-/// The MCP protocol version Luxe speaks, on both transports and the OAuth probe.
+/// The MCP protocol version luxe speaks, on both transports and the OAuth probe.
pub const MCP_PROTOCOL_VERSION: &str = "2025-06-18";
/// Expands `${VAR}` and `$env:VAR` from the environment, left to right, whichever marker comes
diff --git a/crates/luxe-core/src/mcp_config.rs b/crates/luxe-core/src/mcp_config.rs
index 9b8a28d..2ae836f 100644
--- a/crates/luxe-core/src/mcp_config.rs
+++ b/crates/luxe-core/src/mcp_config.rs
@@ -92,7 +92,7 @@ struct StandardServer {
enabled: bool,
#[serde(default)]
timeout_secs: Option,
- /// Luxe's own addition to the shape: other hosts ignore it, and without it an OAuth server
+ /// luxe's own addition to the shape: other hosts ignore it, and without it an OAuth server
/// written the way `docs/guide/mcp.md` shows lost its auth without a word
#[serde(default)]
auth: Option,
@@ -102,7 +102,7 @@ fn default_true() -> bool {
true
}
-/// Read a config file in either the Luxe (`servers`) or standard (`mcpServers`) shape. A missing,
+/// Read a config file in either the luxe (`servers`) or standard (`mcpServers`) shape. A missing,
/// empty, or unparseable file yields no servers (so a stray file never blocks launch), and says so.
fn load_any(path: &Path) -> Vec {
let Ok(text) = std::fs::read_to_string(path) else { return Vec::new() };
@@ -117,8 +117,8 @@ fn load_any(path: &Path) -> Vec {
/// The servers `text` names in either shape, or why it names none it meant to.
///
-/// A Luxe-shape file that fails to parse also reads as an empty standard one, so when the
-/// standard shape finds nothing the Luxe error is the one that explains the file.
+/// A luxe-shape file that fails to parse also reads as an empty standard one, so when the
+/// standard shape finds nothing the luxe error is the one that explains the file.
fn parse_any(text: &str) -> Result, String> {
let luxe = serde_json::from_str::(text);
if let Ok(file) = &luxe {
@@ -224,11 +224,11 @@ impl McpConfigStore {
/// precedence (project overrides user-global; luxe-owned overrides the shared standard file):
///
/// 1. `~/.config/mcp/mcp.json` (standard, user-global) — lowest
- /// 2. `~/.config/luxe/mcp.json` (Luxe, user-global)
+ /// 2. `~/.config/luxe/mcp.json` (luxe, user-global)
/// 3. `.mcp.json` (standard, project)
- /// 4. `.luxe/mcp.json` (Luxe, project) — highest
+ /// 4. `.luxe/mcp.json` (luxe, project) — highest
///
- /// Both the Luxe shape (`{ "servers": [...] }`) and the standard shape
+ /// Both the luxe shape (`{ "servers": [...] }`) and the standard shape
/// (`{ "mcpServers": { id: {...} } }`, as written by Cursor/Claude/Codex) are understood, so an
/// existing MCP setup works with no conversion.
pub fn resolved(repo_root: &Path) -> Vec {
@@ -266,7 +266,7 @@ impl McpConfigStore {
[repo_root.join(".mcp.json"), repo_root.join(".luxe").join("mcp.json")]
}
- /// The shared `mcp/mcp.json` other clients read, then Luxe's own.
+ /// The shared `mcp/mcp.json` other clients read, then luxe's own.
fn user_sources() -> Vec {
let shared = crate::user_dirs::config_home().map(|home| home.join("mcp").join("mcp.json"));
let own = crate::user_dirs::luxe_config_home().map(|luxe| luxe.join("mcp.json"));
diff --git a/crates/luxe-core/src/mcp_config/tests.rs b/crates/luxe-core/src/mcp_config/tests.rs
index 81a99bc..8246010 100644
--- a/crates/luxe-core/src/mcp_config/tests.rs
+++ b/crates/luxe-core/src/mcp_config/tests.rs
@@ -85,7 +85,7 @@ fn resolved_merges_sources_with_project_luxe_winning() {
r#"{ "mcpServers": { "shared": { "command": "a" }, "only-std": { "command": "b" } } }"#,
)
.unwrap();
- // Luxe project file overrides `shared` and adds `only-luxe`.
+ // luxe project file overrides `shared` and adds `only-luxe`.
std::fs::create_dir_all(root.join(".luxe")).unwrap();
McpConfigStore::for_repo(root).add(stdio("only-luxe")).unwrap();
let mut shared = stdio("shared");
@@ -101,7 +101,7 @@ fn resolved_merges_sources_with_project_luxe_winning() {
resolved.sort_by(|a, b| a.id.cmp(&b.id));
let ids: Vec<&str> = resolved.iter().map(|s| s.id.as_str()).collect();
assert_eq!(ids, ["only-luxe", "only-std", "shared"]);
- // The Luxe project entry wins for `shared`.
+ // The luxe project entry wins for `shared`.
let shared = resolved.iter().find(|s| s.id == "shared").unwrap();
assert_eq!(shared.name, "overridden");
}
@@ -109,7 +109,7 @@ fn resolved_merges_sources_with_project_luxe_winning() {
/// A cloned repository's `.mcp.json` must not choose what runs on a run that refuses the
/// project's own tooling.
///
-/// An `eager` server in one used to spawn the moment Luxe opened the tree, headless included,
+/// An `eager` server in one used to spawn the moment luxe opened the tree, headless included,
/// with nothing asked — the same hole `.luxe/hooks.json` is closed against.
#[test]
fn a_refused_project_cannot_start_its_own_servers() {
@@ -236,8 +236,8 @@ fn the_standard_shape_keeps_its_auth() {
assert_eq!(oauth.redirect_uri.as_deref(), Some("http://127.0.0.1:8765/callback"));
}
-/// `"kind": "oauth"` is the spelling in every example, and it parses in Luxe's own shape; the
-/// `o_auth` Luxe used to write still does; and it is written back as `oauth`. Pinned against
+/// `"kind": "oauth"` is the spelling in every example, and it parses in luxe's own shape; the
+/// `o_auth` luxe used to write still does; and it is written back as `oauth`. Pinned against
/// the literal, because a round trip through the same enum passes whatever the spelling is.
#[test]
fn oauth_is_spelled_oauth() {
@@ -287,7 +287,7 @@ fn a_secret_is_hidden_from_debug_and_survives_a_rewrite() {
);
}
-/// A file Luxe creates is owner-only, since it can hold a secret; a mode the user chose stays.
+/// A file luxe creates is owner-only, since it can hold a secret; a mode the user chose stays.
#[cfg(unix)]
#[test]
fn saving_starts_private_and_keeps_a_chosen_mode() {
@@ -302,9 +302,9 @@ fn saving_starts_private_and_keeps_a_chosen_mode() {
assert_eq!(mode(store.path()), 0o644, "a mode the user set is theirs");
}
-/// A Luxe-shape file with one entry that does not parse is reported, not read as no servers.
+/// A luxe-shape file with one entry that does not parse is reported, not read as no servers.
///
-/// The break this catches: the failed Luxe parse fell through to the standard shape, which read
+/// The break this catches: the failed luxe parse fell through to the standard shape, which read
/// the same file as an empty `mcpServers` map, so every server in it vanished without a log line.
#[test]
fn a_luxe_file_with_one_bad_entry_is_reported() {
diff --git a/crates/luxe-core/src/mcp_stdio.rs b/crates/luxe-core/src/mcp_stdio.rs
index 249c155..4229a0f 100644
--- a/crates/luxe-core/src/mcp_stdio.rs
+++ b/crates/luxe-core/src/mcp_stdio.rs
@@ -171,7 +171,7 @@ impl McpStdioClient {
impl Drop for McpStdioClient {
/// `kill_on_drop` reaches only the server. A server that delegates spawns work of its own —
- /// a headless Luxe, a browser — and that went on running, writing files, after the session
+ /// a headless luxe, a browser — and that went on running, writing files, after the session
/// that asked for it had exited.
fn drop(&mut self) {
self.kill_group_if_running();
diff --git a/crates/luxe-core/src/mcp_stdio/tests.rs b/crates/luxe-core/src/mcp_stdio/tests.rs
index f956548..5fba0a6 100644
--- a/crates/luxe-core/src/mcp_stdio/tests.rs
+++ b/crates/luxe-core/src/mcp_stdio/tests.rs
@@ -108,7 +108,7 @@ for line in sys.stdin:
std::env::set_var("LUXE_TEST_REAL_TOKEN", "supersecret");
let mut env = std::collections::HashMap::new();
- // The value uses `${VAR}` interpolation against Luxe's env.
+ // The value uses `${VAR}` interpolation against luxe's env.
env.insert("LUXE_TEST_TOKEN".to_string(), "${LUXE_TEST_REAL_TOKEN}".to_string());
let config = McpServerConfig {
@@ -236,7 +236,7 @@ for line in sys.stdin:
assert!(error.to_string().len() < 30_000, "a bounded tail: {}", error.to_string().len());
}
-/// `initialize` offers the protocol version Luxe speaks.
+/// `initialize` offers the protocol version luxe speaks.
#[tokio::test]
async fn initialize_offers_the_current_protocol_version() {
if std::process::Command::new("python3").arg("--version").output().is_err() {
diff --git a/crates/luxe-core/src/mention_source.rs b/crates/luxe-core/src/mention_source.rs
index a5fa94e..e7bc5f3 100644
--- a/crates/luxe-core/src/mention_source.rs
+++ b/crates/luxe-core/src/mention_source.rs
@@ -1,6 +1,6 @@
//! Mention sources — a program the `@` picker lists references from.
//!
-//! Luxe knows files and GitHub. Everything else a person works from — Jira issues, Asana tasks, a
+//! luxe knows files and GitHub. Everything else a person works from — Jira issues, Asana tasks, a
//! runbook index — is a program they point it at: it prints one JSON object a line, and each line
//! becomes a row in a tab of its own. A picked row goes into the composer as `@[kind:id]` and
//! reaches the model as the row's `text`, a reference and never the thing itself; the agent looks
diff --git a/crates/luxe-core/src/oauth_loopback.rs b/crates/luxe-core/src/oauth_loopback.rs
index edae920..c292e74 100644
--- a/crates/luxe-core/src/oauth_loopback.rs
+++ b/crates/luxe-core/src/oauth_loopback.rs
@@ -280,7 +280,7 @@ pub fn page(title: &str, message: &str, accent: &str) -> String {
format!(
" \
\
- Luxe
-Pair this device with Luxe
+Pair this device with luxe