post-change-check
Skill2 starsChanged 14 days agoMITView on GitHub
Orchestrates the end-of-turn review after non-trivial edits to app/, components/, lib/, scripts/, or config files. Runs the docs-sync audit, invokes the code-review and quality-check skills on the diff, and flags anything still missing before the turn can close.
- Reads credentials
In this repository5 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- agent-friendlySkill
- code-reviewSkill
- quality-checkSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
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
- Distinctive: line 8
Run this when a turn has made code changes and you're about to report done.
- Distinctive: line 26
Keep a mental list of: touched files, added/removed exports, new deps, new env vars, new routes.
- Very specific: line 36
Every item must point to a real tasks/X.Y.Z/NN-*.md file — no dangling taskFile references.
- Distinctive: line 98
If every phase is clean, say so plainly in one line.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Post-change check — docs sync + code review + quality sweep
Run this when a turn has made code changes and you're about to report done. It's the one stop that verifies the work is coherent (docs and code aligned) and correct (conventions followed).
When to run
- Auto-invoke yourself when a turn edits any of:
app/**,components/**,lib/**,scripts/**,.claude/**,lefthook.yml,biome.json,package.json,tsconfig.json,next.config.ts. - User can also invoke manually:
/post-change-check. - Skip for pure typo fixes, comment tweaks, or changes entirely inside
tasks/anddata/.
How to run
Work the four phases in order. Don't skip ahead — the docs phase often surfaces issues code review won't.
Phase 1 — Collect the diff
git status --short— see what moved.git diff --stat HEAD— scope.git diff HEAD -- app/ components/ lib/ scripts/— the actual changes.
Keep a mental list of: touched files, added/removed exports, new deps, new env vars, new routes.
Phase 2 — Docs sync audit
For each category below, decide yes / no / N/A, then act on the yeses:
- Changelog line (rule: AGENTS.md → Versions, changelog, roadmap)
- Yes only for a user-facing capability. No for hygiene, contributor docs, refactors, dep bumps or UI polish without a semantic change.
- Action: add a one-line bullet under "Shipped since …, owed a changelog line" in the next version's
tasks/<version>/README.md(lib/changelog.tsgets it when the release is cut), and delete any shipped item fromlib/roadmap.ts.
lib/roadmap.tsintegrity- Every item must point to a real
tasks/X.Y.Z/NN-*.mdfile — no danglingtaskFilereferences. - Adding a roadmap entry requires adding its task file in the same PR. Don't split.
- Released versions don't live on the roadmap; they move to
lib/changelog.tswith a release label.
- Every item must point to a real
tasks/X.Y.Z/files- Yes if scope shifted, a task was partly/fully completed, or a new sub-task emerged.
- Action: update the file's
**Status**:header (planned/in_progress/done) and its "What shipped" section.
AGENTS.md- Yes if stack, directory layout, conventions, or I/O boundary changed.
- Specifics:
- Stack changes (Next.js / React / Tailwind / DB driver) → update Stack.
- New top-level folder or
lib/area → update the Layout tree (it lists directories, not every file). - New page in
app/→ reachable from the nav, footer or a parent page, and inapp/sitemap.tsunless noindex. - New signal / model / host → ensure the matching "Adding a …" section in
AGENTS.mdis still accurate. - New convention or constraint → add a bullet under Conventions.
README.md- Yes if a product-facing behaviour, command, or screenshot changed.
- Specifics: "Run locally" commands must work; the agent list, signal count and Public API section must match the code.
CONTRIBUTING.md- Yes if the branch/commit/PR flow or review bar changed.
.env.example- Yes if a new env var was introduced.
- Action: document with a one-line comment above the line.
.claude/skills/*- Yes if a new agent workflow was added or an existing one changed.
- Action: update or create a skill file.
Report each as ✓ updated, ✓ N/A — reason, or ✗ missing — file:line where it should go.
Phase 3 — Code review
Invoke the code-review skill on the diff and report in its format.
Phase 4 — Quality sweep
Invoke the quality-check skill on the diff and report in its format.
Output format
Print a single consolidated report:
Post-change check — <N> files touched
Docs sync
✓ tasks/0.10.0/README.md — owed-changelog bullet added
✓ N/A — README.md (no product-facing change)
✗ tasks/0.2.0/01-tests.md — status still "planned" but harness already committed; flip to in_progress
Code review
Must fix: <none> | <file:line — why>
Should fix: …
Consider: …
Quality
a11y: pass
responsive: pass
performance: pass — flagged `new-dep@1.2.3` at 73 KB gz, justify or swap
security: pass
Next steps: <bullet list or "clean, ready to commit">If every phase is clean, say so plainly in one line. Don't invent findings to look thorough.
What this skill does NOT do
- It does not run tests, biome, or tsc — those are the pre-commit hook's job (
lefthook.yml). - It does not push, commit, or open PRs.
- It does not modify the codebase beyond the docs updates in Phase 2 (tasks / roadmap / AGENTS / README).
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
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.

