agentleFS
vosslabvosslab/vosslab-skills/skills/documentation/agents-md-fixer/SKILL.md

agents-md-fixer

Skill5 starsChanged 21 days agoMITView on GitHub

Trim bloated `AGENTS.md` files into concise pointers to canonical `docs/*.md` rules. Use when agent guidance duplicates repo documentation. This skill changes only `AGENTS.md`; use documentation skills for README or `docs/` changes.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    Use the docs/REPO_STYLE.md AGENTS.md files policy for size and structure.

    line 17
  2. Distinctive:

    Keep only items that are repo-specific and load-bearing:

    line 21
  3. Distinctive:

    Test invocation cheat-line specific to this repo.

    line 33
  4. Very specific:

    Commit-message and changelog rules already in docs/CHANGELOG.md or docs/REPO_STYLE.md.

    line 48
  5. Distinctive:

    If this line were removed today, would an agent's behavior change on a real task?

    line 61
  6. Very specific:

    Read the current AGENTS.md and the docs/ index.

    line 69
  7. Very specific:

    Check wc -l AGENTS.md against the repository policy.

    line 70
  8. Very specific:

    No duplication of canonical docs/*.md content.

    line 96
  9. Distinctive:

    Every external concept is referenced by path, not restated.

    line 97
  10. Distinctive:

    No deletion of genuinely repo-specific operational rules.

    line 98
  11. Distinctive:

    No new sections with content that has no enforcement or no concrete action.

    line 99

≈ “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.

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

AGENTS.md fixer

Goal

Make AGENTS.md a tiny pointer file. It should mostly be bare paths into docs/*.md, not prose. This skill audits and tightly rewrites it using the repository's load-bearing rules.

AGENTS.md is agent-facing. Reference docs by bare path (docs/PYTHON_STYLE.md), not Markdown links.

Use the docs/REPO_STYLE.md AGENTS.md files policy for size and structure.

Required content (keep)

Keep only items that are repo-specific and load-bearing:

  • Bare path pointers to canonical style docs that exist in this repo, for example docs/REPO_STYLE.md, docs/PYTHON_STYLE.md, docs/MARKDOWN_STYLE.md, docs/PYTEST_STYLE.md, and docs/CHANGELOG.md.
  • Repo-specific runtime quirks: interpreter bootstrap, env activation, container entry, non-default install paths.
  • Non-obvious workflow constraints actually enforced in this repo (example: "humans run git commit", branch policy, hook behavior).
  • Test invocation cheat-line specific to this repo.
  • Any user-injected directive marked as overriding defaults (preserve it verbatim).

Frivolous content (cut or move)

Flag and remove these patterns:

  • Long philosophy or "why we believe X" paragraphs. Move the canonical version to docs/REPO_STYLE.md, or delete if already covered there.
  • Restated style rules already in docs/PYTHON_STYLE.md, docs/MARKDOWN_STYLE.md, or docs/REPO_STYLE.md.
  • Aspirational guidance with no enforcement and no concrete action.
  • Generic AI etiquette unrelated to this repo (tone, manners, hedging advice).
  • Commit-message and changelog rules already in docs/CHANGELOG.md or docs/REPO_STYLE.md.
  • "Future considerations" sections with no owner and no date.

Manager judgment rubric

For each line or section, answer in order. Stop at the first "yes" that demands action.

  1. Is this repo-specific (would it differ in a sibling repo)? If no, cut or replace with a link to the canonical doc.
  2. Is the same rule already written in a docs/*.md file? If yes, replace with a link to that file.
  3. If this line were removed today, would an agent's behavior change on a real task? If no, cut.
  4. Is the canonical home a sibling doc that already exists? If yes, move the content there and link from AGENTS.md.
  5. Otherwise, keep, and trim wording to a single short bullet.

Workflow

  1. Read the current AGENTS.md and the docs/ index.
  2. Check wc -l AGENTS.md against the repository policy. If it is already a tight pointer file and passes the rubric, report no action needed and stop.
  3. For each section, apply the rubric. Categorize each as one of: keep, link-only, move-to-docs, or delete.
  4. For move-to-docs items, identify the right docs/*.md target using the recommended common docs list in docs/REPO_STYLE.md. If the target doc does not exist yet, do not create it here. Note it as a follow-up for docset-updater, setup-install-usage-docs, or arch-docs.
  5. Rewrite AGENTS.md minimally with short sentence-case headings and - bullets.
  6. Preserve any user instruction explicitly marked as overriding defaults.

Markdown house rules

  • ASCII only; escape symbols like α if needed.
  • Sentence case headings, short headings (3-6 words).
  • Bullets use -, one idea per bullet.
  • Reference docs by bare path, optionally in backticks; do not use Markdown link syntax.

Quality bar

  • Final AGENTS.md is small: prefer about 15 non-blank lines, hard cap about 50. Smaller is always better. If you cannot get under 50, more content belongs in docs/*.md.
  • Most of the file is bare path pointers into docs/*.md, not prose.
  • No duplication of canonical docs/*.md content.
  • Every external concept is referenced by path, not restated.
  • No deletion of genuinely repo-specific operational rules.
  • No new sections with content that has no enforcement or no concrete action.

Inputs to inspect

  • Current AGENTS.md content.
  • The list of files under docs/ (so the skill can choose link targets that actually exist).
  • Any standing override already marked for verbatim preservation.

Output

  • The applied AGENTS.md patch with minimal edits.
  • A change log with three short lists:
    • moved: each chunk moved into docs/*.md, with the destination file.
    • linked: each chunk replaced by a bare path pointer, with the target.
    • deleted: each chunk deleted, with one sentence of justification.
  • A short follow-up list of docs/*.md stubs that should exist, naming the skill that owns each one (docset-updater, setup-install-usage-docs, arch-docs, or readme-docs).

References

  • docs/REPO_STYLE.md: "AGENTS.md files" section sets the 100-150 line target and the "concise and operational" rule.
  • docs/MARKDOWN_STYLE.md: heading and link conventions used in the rewrite.
  • ../readme-docs/SKILL.md: sibling single-artifact standardizer with the same shape.

In a docset refresh, run after the linked docs/*.md files exist and alongside screenshot-docs; this skill owns only AGENTS.md.

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.