agentleFS
oldmilkyoldmilky/luno-for-cc/.claude/skills/check/SKILL.md

check

Skill0 starsChanged 4 months agoMITView on GitHub

Fast read-only review of the uncommitted working tree. Use when the user says /check, or mid-feature before continuing or committing. Cheaper than a full review and never edits anything.

87 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use daily-arxiv, exp-eval and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • JuliusBrusseeSkill

    backprop

    Bug → spec protocol. When a bug is found or a test fails, trace the cause, decide whether a new §V invariant would catch recurrence, append to §B. This is the one non-obvious thing SDD does that plan-then-execute doesn't. Triggers on test failure, bug report, post-mortem, or explicit user ask.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-eval

    Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-run

    Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.

    Also in 4 projects that use this

  • +6 more from Lambenthan
  • +1 more from skyllwt

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 74 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use it mid-feature after a meaningful chunk of work, and before /gitpush as a safety net.

    line 11
  2. Distinctive:

    Every HIGH and CRITICAL gets one skeptic whose job is to refute it: read the code around the claim and say whether the described failure …

    line 53
  3. Distinctive:

    Read-only: No edits during the review.

    line 90

“Common” is measured: kept by at least 6 in 10 of the 74 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

A gut-check between edits: run the gates, then fan the specialist reviewers over uncommitted changes only. Read-only — no edits, no commit, no push.

Use it mid-feature after a meaningful chunk of work, and before /gitpush as a safety net. It is not a branch review: committed history is out of scope.

Phase 0 — Scope and gates

git status --porcelain
git diff --name-only
git diff --staged --name-only

Empty on all three → say there is nothing uncommitted and stop.

Then run the gates, because a review of code that does not compile is wasted:

bun run lint
bun run test

Report a failure immediately and stop — a red gate is the finding.

Phase 1 — Fan out

Split the changed files and dispatch in one message so they run in parallel:

ChangedReviewer
anything under src/**extension-security-reviewer
anything under webview/src/**webview-design-reviewer
eithera correctness pass — see below

Give each agent the explicit file list and this instruction: review only these files, against the working-tree diff, and report HIGH/CRITICAL first.

The correctness pass is a general-purpose agent asked to find bugs a reader would miss: wrong condition, off-by-one, a promise not awaited, an error path that swallows, state that can go stale. Tell it to ignore style — the linters own that now.

Phase 2 — Verify before reporting

Every HIGH and CRITICAL gets one skeptic whose job is to refute it: read the code around the claim and say whether the described failure can actually happen, with a concrete input or sequence. Default to refuted when unsure.

Findings that survive get reported. Refuted ones get one line each. This step is not ceremony — in past rounds it caught a motion.li conversion that killed a class-level opacity, an overflow: hidden clipping a focus ring, and a truncation bug inside the tooltip primitive itself.

MEDIUM and LOW are listed unverified, one line each. Do not expand them.

Phase 3 — Report

## /check — working tree (<n> files)

Gates: lint <state> · test <n> passed / <n> skipped

### Confirmed HIGH / CRITICAL (<n>)
1. **<title>** — [<file>:<line>](<file>#L<line>)
   <what breaks, and the input or sequence that breaks it>
   Fix: <concrete>

### Worth a glance (<n>)
- <title> — [<file>](<file>)

### Refuted (<n>)
- <title> — <one line why not>

### Verdict
clean, keep going  /  <n> to fix before commit

Nothing confirmed → a single line: /check clean across <n> files, gates green.

Hard rules

  • Read-only. No edits during the review. Offer to fix afterwards as a separate, explicit step.
  • Working tree only. Committed history belongs to a branch review.
  • Only HIGH/CRITICAL are verified. Verifying everything turns a gut-check into the thing it exists to avoid.
  • No lint-level noise. eslint, stylelint and prettier already ran; repeating them is not a finding.
  • A claim about rendered behaviour needs /browser, not an assertion.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.