agentleFS
Cody-SimsCody-Sims/agent-skills/skills/documentation-maintenance/SKILL.md

documentation-maintenance

Skill0 starsChanged 4 months agoMITView on GitHub

Updates README, API docs, architecture notes, ADRs, and changelogs to match current code. Use when a code/configuration change invalidates documentation, or when writing an ADR or changelog entry; not for new requirements or maintaining a .shadow decision graph.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Read applicable instructions and the change.

    line 21
  2. Distinctive:

    Search likely documentation paths for changed names, commands, flags, paths, and values, including removed/renamed forms.

    line 24
  3. Distinctive:

    Update, supersede, or delete stale passages instead of appending corrections.

    line 28
  4. Distinctive:

    Check affected documented setup sequences and commands by running them safely or tracing each step exactly against current code.

    line 43
  5. Distinctive:

    Read updated documents for internal consistency.

    line 48
  6. Very specific:

    Stop when affected documents agree with code and applicable checks are accounted for.

    line 56
  7. Distinctive:

    If a command or example cannot be checked, state the limitation and leave that claim unverified.

    line 59

“Common” is measured: kept by at least 6 in 10 of the 7 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

Documentation Maintenance

Contract

Inputs: the change or requested document, current implementation, and repository documentation conventions. Update affected documentation, not application behavior. Create only requested or change-required artifacts.

Workflow

  1. Read applicable instructions and the change. Identify affected README setup and commands, API signatures/errors/examples, architecture responsibilities and diagrams, ADRs, changelog, and nearby comments/docstrings.
  2. Search likely documentation paths for changed names, commands, flags, paths, and values, including removed/renamed forms. Batch related searches and widen only where references lead. Resolve every identified contradiction; do not launch a worker or audit unrelated architecture for a small documentation fix.
  3. Update, supersede, or delete stale passages instead of appending corrections. Explain relevant rationale and constraints, use terminology consistent with code, and use relative repository links. Verify claims against implementation, not an outdated change summary.
  4. Handle records only when applicable:
    • ADR: for a significant, hard-to-undo decision or reversal, read the ADR template. Record context/forces, decision, rejected alternatives, consequences/trade-offs, stable number, date, and status (proposed, accepted, superseded, deprecated). Do not invent acceptance or rewrite an accepted decision; supersede with a new linked ADR.
    • Changelog: for user-visible changes, read the entry format and add an unreleased entry using existing conventions. Classify as Added, Changed, Deprecated, Removed, Fixed, or Security; describe user impact and breaking-change migration. Skip purely internal, behavior-preserving changes unless policy requires it.
  5. Check affected documented setup sequences and commands by running them safely or tracing each step exactly against current code. Compile/run affected code examples when possible. Record which checks ran versus were only traced; never claim an unexecuted example runs. Respect authorization and environment limits rather than running destructive setup or installing speculatively.
  6. Read updated documents for internal consistency. For long documents, inspect edited sections, headings, and affected cross-references rather than rereading unrelated prose. Check relative links and search the affected documentation for stale identifiers. Run repository-required documentation gates; otherwise use existing focused link/example checks, not unrelated full application tests.

Stop and output

Stop when affected documents agree with code and applicable checks are accounted for. Report paths updated/superseded/deleted, ADR status changes, changelog entry, commands/examples and actual results, and unresolved documentation gaps. If a command or example cannot be checked, state the limitation and leave that claim unverified; do not silently declare it runnable.

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.