agentleFS
lukesw55lukesw55/ai-pm-toolkit/.claude/skills/repo-doctor/SKILL.md

repo-doctor

Skill1 starsChanged 42 days agoMITView on GitHub

Read-only health check for this workspace. Use before committing changes under skills/, hooks/, .claude/, or .codex/, when validating skills/hooks/memory consistency, or when the user asks "valida o repo", "checa as skills", "tem ref quebrada?", "algum binário grande?", "repo lint", "repo-doctor". Verifies stage aliases, memory skeleton, real path references, large files (working tree + git history), hooks cited in docs, skill frontmatter, and tracked-vs-ignored local artifacts. Reports findings only — never edits, deletes, or touches the remote.

  • Commits and pushes

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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
  1. Distinctive:

    Read-only consistency check for this PM workspace.

    line 8
  2. Distinctive:

    Run each from the repo root.

    line 12
  3. Very specific:

    Read-only. Never Edit/Write/rm, never git push/filter-repo/rebase, never delete memory.

    line 26

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.

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

repo-doctor

Read-only consistency check for this PM workspace. Run the checks below, then emit the findings table. Never edit, delete, or run anything that touches the git remote.

Checks

Run each from the repo root. Treat any surprising result as a finding.

  1. Stage schema — python3 scripts/validate_context.py (exit 0 = ok) and python3 scripts/advance_stage.py --list. Confirm active-context.md resolves to a canonical stage; flag a raw discover that is not being aliased to discovery. 1b. Memory caps — python3 scripts/memory.py doctor (exit 0 = ok). Flags: pointer over 2 KB, more or fewer than one ACTIVE block, missing Current stage:, parked slugs without a projects/ dir or state.md, changelogs holding more than 3 entries, warm files (changelog.md, session-kickoff.md, state.md, profile.md, decisions.md, insights.md) and the shared org files (.ai/memory/org/*.md) over their soft caps (distill candidates, or prose to trim), pending .distill/ packages, a stale archive index (run memory.py index <slug>), and PII-named project dirs the scripts skip. It also warns on e-mail and phone patterns inside .ai/memory/org/*.md: a role alias is fine, a person is not, and a name is a review rule no pattern catches.
  2. Memory skeleton versioned — git status --short --ignored .ai/memory should show _templates/* and active-context.example.md as versionable while real memory stays ignored. Confirm with git check-ignore -v .ai/memory/active-context.md .ai/memory/projects .ai/memory/inbox.md .ai/memory/index.md .ai/memory/org (no real memory leaking into git; .ai/memory/org is ignored upstream and un-ignored only in a fork).
  3. Real path references — for each references/...md or sibling-relative ../<skill>/...md path mentioned in any SKILL.md under skills/, confirm the target file exists relative to the file that cites it (not a fixed depth — a SKILL.md and a references/*.md file resolve ../ differently). Resolve the path before flagging (a sibling skill's references/ dir is valid). This is the corrected check: do not flag a skill just because its own folder lacks a references/ dir. python3 scripts/validate_repo.py automates the backtick and markdown-link forms of this check (check_backtick_paths, check_markdown_links); the manual pass here covers what the regex skips — bare filenames and paths embedded in prose.
  4. Large files — working tree: find . -type f -size +5M -not -path './.git/*'. History: git rev-list --objects --all | git cat-file --batch-check='%(objecttype) %(objectsize) %(rest)' | awk '/^blob/{print $2,$3}' | sort -rn | head. Flag big blobs and note whether they are in the working tree, history, or both.
  5. Hooks cited vs present — for every *.sh named in CLAUDE.md, AGENTS.md, .claude/settings*.json, and .codex/hooks.json, confirm the file exists under hooks/.
  6. Skill frontmatter — every skills/*/SKILL.md has a YAML frontmatter block with name and description.
  7. Local artifacts ignored — git status --short shows no large untracked junk; git check-ignore covers *.deb, **/node_modules/, and .ai/memory/projects/**/.browser-profile/.
  8. Mirror parity — python3 scripts/sync_skills.py --check exits 0; skills/, .claude/skills/, and .agents/skills/ are byte-identical (excluding workspace/).

Guardrails

  • Read-only. Never Edit/Write/rm, never git push/filter-repo/rebase, never delete memory.
  • Cite a path or the exact command output for every finding.
  • Suggest a fix; do not apply it.

Output contract

| Severity | Area | Finding | Path / command | Suggested fix |
|----------|------|---------|----------------|---------------|

Severity = Alta / Média / Baixa. End with a one-line verdict: clean, or N findings (M Alta).

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.