AGENT SKILL

project-check

Bring a workspace to a state where the full gate runs, and interpret what it reports.

License 许可
Apache-2.0
最近更新
2026-11-10
Stars
★ 0
Forks
0

01 前言 Front Matter

name
project-check
description
Bring a workspace to a state where the full gate runs, and interpret what it reports.

03 SKILL.md 正文

Run the full gate

  1. Read AGENTS.md. The Justfile is the only supported interface to the checks; do not assemble an equivalent pipeline by hand.
  2. Check the prerequisites exist: rustup's cargo first on PATH (just check-toolchain says so), the pixi environment at .pixi/envs/default/bin (nextest, llvm-cov, deny, shear, taplo, prek, just, the bg-* scripts), .tools/bin/cargo-hack, .tools/bin/rustqual (built from source by just install-tools), the pinned checker at .tools/bin/allium, Cargo.lock and pixi.lock. If any is missing, just initialize creates them — it normalises formatting and installs the hook, and it never stages, commits, tags or pushes — but it downloads, so it is a network operation and needs explicit authorization before it runs, unless the task already grants it (a ticket's Prepare step does). The checker alone is just install-allium and rustqual alone is just install-tools, under the same rule; both are gitignored and per-worktree, so a fresh worktree needs the checker before gates 3, 17 and 18 can pass, and rustqual before gate 7 can.
  3. Run just check. It runs each recipe in order and snapshots the worktree between them.
  4. Read only the first failure. The gates are ordered so that a later failure is often a consequence of an earlier one.
  5. A report that a recipe changed the worktree is a defect in that recipe, not in the change under test. Checks are read-only; just fix is where mutation belongs.
  6. Hand a failing gate to the fix-quality skill rather than working around it.
  7. Confirm with just check-clean that the worktree is unchanged before reporting success.

04 同仓库更多 Skills

steven-cutting
steven-cutting/libpawdoku

witness

Independently witness that an Allium loop's convergence claim is true and was reached honestly. Use when the user wants to verify a loop's self-report, confirm tests really pass and no generated test was weakened, produce a convergence certificate or witness record, gate CI on a trustworthy signal, or check that an autonomous run did not cheat its way to green.

★ 0
steven-cutting
steven-cutting/libpawdoku

weed

Weed the Allium garden. Find where Allium specifications and implementation code have diverged, and help resolve the divergences. Use when the user wants to check spec-code alignment, compare specs against implementation, audit for spec drift or violations, sync specs with code or code with specs, or verify whether the implementation matches what the spec says.

★ 0
steven-cutting
steven-cutting/libpawdoku

tend

Tend the Allium garden. Use when the user wants to write, edit, update, add to, improve, clarify, refine, restructure, fix or migrate Allium specs. Covers adding entities, rules, triggers, surfaces and contracts, fixing syntax or validation errors, renaming or refactoring within specs, migrating specs to a new language version, and translating requirements into well-formed specifications. Pushes back on vague requirements.

★ 0
steven-cutting
steven-cutting/libpawdoku

spec-change

Change an Allium specification and carry the change through the tests and the implementation.

★ 0
steven-cutting
steven-cutting/libpawdoku

rust-change

Implement or review a change to the engine with the trait boundary, no_std, tests and documentation evidence the invariants require.

★ 0
steven-cutting
steven-cutting/libpawdoku

review-docs

Add or revise a handbook page so it satisfies the documentation contract and stays reachable.

★ 0