agentleFS
MandalorianBatmanMandalorianBatman/austere-axi/.agents/skills/review-change/SKILL.md

review-change

Skill8 starsChanged 4 days agoMITView on GitHub

Test and review an austere-axi change before it is committed or merged — runs the mechanical checks, launches the change-reviewer, visual-qa and cold-start-author agents as needed, and triages their findings. Maintainer-only (not shipped).

32 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use plan-change, implement-change and 8 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.
  • BanKnightSkill

    plan-change

    将已完成 spec/design 的 change 拆成可执行 plan 与 tasks。用户要把设计落成执行计划、任务清单或实现步骤时使用。

    Also in 7 projects that use this

  • Codagent-AISkill

    implement-change

    Implements an approved change in one continuous session, with tiered test runs. Use when asked to implement a change from its approved definition, or to invoke codagent:implement-change.

    Also in 7 projects that use this

  • MaheshAwasareSkill

    write-runbook

    Generate an operational runbook for a service from its code, config, and dependencies — what to do when it breaks, where to look, how to roll back, who to escalate to. Use proactively before a service goes to production, after a major incident, or when handing off operational ownership.

    Also in 3 projects that use this

  • Sagargupta16Skill

    upgrade-dependencies

    Upgrade a project's dependencies one at a time, running the test suite after each bump and reverting the ones that break it. Use it during routine dependency maintenance, when a batch of updates has accumulated; before a major release, to go into it on current and patched dependencies; when Renovate or Dependabot PRs have piled up and you want them resolved locally in one sitting; when a security advisory means you need the patched version in, and you need to know what it breaks.

    Also in 3 projects that use this

  • testified-ossSkill

    incident-triage

    Triage production incidents with severity assessment, root cause hypothesis, impact analysis, and escalation recommendations. Use when analyzing outages, errors, or production issues.

    Also in 4 projects that use this

  • srclightSkill

    refactor-safely

    This skill should be used when the user asks to "refactor", "rename", "move", "change a function signature", "extract a method", "delete unused code", "clean up", or any modification that could break callers or dependents. Provides a pre-change safety checklist using srclight's graph tools to assess blast radius before touching code.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 27 edited versions we read keep these.

  • 5

    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. Distinctive:

    Use this skill before committing a non-trivial change, or when asked to review a branch.

    line 11
  2. Distinctive:

    Leave out your own assessment, what you think is fine, the commit message's claims, and why you made each decision.

    line 64
  3. Very specific:

    Other harnesses: start a fresh session or subagent per agent, give it the body of .agents/agents/<name>.md plus its inputs, and restrict it …

    line 69
  4. Distinctive:

    Disagreements: if change-reviewer and visual-qa disagree, the rules in AGENTS.md and the add-component skill decide.

    line 83
  5. Distinctive:

    After fixes, re-run the mechanical checks.

    line 85
  6. Distinctive:

    Write one short block for the user, and put the same summary in the commit message:

    line 89

“Common” is measured: kept by at least 6 in 10 of the 27 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

Reviewing a change to austere-axi

Use this skill before committing a non-trivial change, or when asked to review a branch.

Mechanical checks run as code, because they are cheap and can't be talked out of a failure. Judgment comes from separate agents. They run in fresh contexts, so they are not anchored on how the change was made. Your job is to scope the change, brief the agents, and decide what to do with their findings.

1. Scope

git merge-base HEAD main      # base, unless you were given one
git diff --stat <base>        # includes uncommitted work

Classify the changed files:

AreaPaths
componentssrc/runtime/components/, packs/, src/runtime/stock.ts, src/runtime/define.ts, test/fixtures/signatures.json
looksrc/runtime/austere.css, src/chrome/
runtimeother src/runtime/, src/pack-kit/
agent-facing textsrc/runtime/prompt.ts, the help text in src/cli/main.ts, component descriptions and notes
cli/serverother src/cli/
docs/toolingeverything else

2. Mechanical checks (always)

npm run build && npm test
npm run test:browser      # when components, look or runtime changed

test:browser needs Chrome, or CHROME_PATH pointing at a browser, and Firefox (npx playwright-core install firefox, or FIREFOX_PATH). If either is missing, say so in the report; do not silently skip it.

Fix every failure before launching any agent. Agents are slow and cost tokens, so don't spend them on a red build.

3. Launch the agents

Launch every relevant agent at once, in parallel. Only launch the ones the change needs:

Changedchange-reviewervisual-qacold-start-author
componentsyesyesyes
lookyesyes
agent-facing textyesyes
runtime, cli/serveryes
docs/tooling only

For a docs- or tooling-only change, read it yourself; no agent is needed.

Give each agent only these inputs:

  • change-reviewer: the base ref, a head ref if you are reviewing a range, and the list of changed files.
  • visual-qa: the changed component names, plus any pages under examples/ that use them.
  • cold-start-author: the repository path, an output path outside the repository (such as /tmp/austere-cold-start/<name>.html, so the harness doesn't load the repo's docs into its context), and a scenario: a realistic user request in which the changed component is the natural choice. Do not name the component. Whether the agent finds it is part of the test. For an agent-facing text change, choose a scenario that exercises the changed instruction.

Leave out your own assessment, what you think is fine, the commit message's claims, and why you made each decision. An agent handed your conclusions tends to return them confirmed.

How to launch:

  • Claude Code: the agents live in .claude/agents/, symlinked to .agents/agents/. Use them as subagent types change-reviewer, visual-qa and cold-start-author.
  • Other harnesses: start a fresh session or subagent per agent, give it the body of .agents/agents/<name>.md plus its inputs, and restrict it to the tools listed in the file's frontmatter where the harness allows.

4. Triage

Agents report confidently whether they are right or wrong. Before acting on a finding, verify it yourself: read the cited line, or reproduce the scenario.

  • Blockers and should-fix: fix them, or decline in writing with the reason (the finding was wrong, or the fix would be out of scope).
  • Nits: fix them if they are cheap, otherwise drop them.
  • cold-start-author:

    • It did not pick the changed component for a scenario that needed it: the description or the group notes failed.
    • It picked the component but misused it: the call design or the description failed.
    • It needed several fix rounds: the first check output tells you which instruction was missing.

    Fix the docs, not the scenario.

  • Disagreements: if change-reviewer and visual-qa disagree, the rules in AGENTS.md and the add-component skill decide. If the rules are silent, decide yourself and record the decision.

After fixes, re-run the mechanical checks. Re-run an agent only when a fix changed what that agent judged (for example, a reworked component design); individual fixes don't need it.

5. Report

Write one short block for the user, and put the same summary in the commit message:

Reviewed:
  tests: npm test ✓, test:browser ✓ (or: not run — no browser)
  change-reviewer: ship after fixes — 1 should-fix fixed, 1 nit declined (reason)
  visual-qa: looks right after 2 fixes
  cold-start-author: picked FileTree unprompted, 0 fix rounds; 1 doc fix

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.