agentleFS
Sequela02Sequela02/canasta/skills/workflow/documentation-hygiene/SKILL.md

documentation-hygiene

Skill0 starsChanged yesterdayMITView on GitHub

Place, consolidate, or retire repository documentation; use before

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep repository documentation useful to its intended readers.

    line 15
  2. Distinctive:

    Before creating or retaining a document:

    line 19
  3. Distinctive:

    Read the repository guidance, documentation entry points, and relevant existing owners.

    line 20
  4. Distinctive:

    Verify current claims against their source owners.

    line 42
  5. Distinctive:

    Update links, indexes, build references, and affected consumers with moves or removals.

    line 59
  6. Very specific:

    Run repository link checks and documentation builds when affected, regenerate generated docs and confirm no diff, run the doc examples' …

    line 62
  7. Very specific:

    Use task-closeout for the rest of closing a task: scratch files, private material, handoffs and branches.

    line 67
  8. Very specific:

    Use technical-writing for prose and page content (its developer-docs reference for library and CLI docs), research for external evidence, …

    line 78

≈ “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

Documentation Hygiene

Keep repository documentation useful to its intended readers. Apply this procedure to the requested scope and the docs the task created or changed; ordinary task completion does not authorize a repository-wide cleanup.

Before creating or retaining a document:

  1. Read the repository guidance, documentation entry points, and relevant existing owners. Identify the reader, purpose, sensitivity, expected lifetime, and canonical destination. Prefer updating the existing owner to creating a parallel explanation. Do not create folders or templates just to fill a taxonomy.
  2. Distinguish maintained documentation, durable decision or evaluation evidence, requested deliverables, and temporary working state. A plan or report is not automatically permanent because an agent produced it. Keep scratch notes, transcripts, intermediate reports, and task bookkeeping in the configured private project state or an appropriate temporary location outside tracked documentation.
  3. Decide whether content belongs in the repository before writing it. Contributor guides, architecture, and public design discussions can be legitimate public docs. Private operational details, personal paths, or session context require an appropriate private destination or a sanitized public explanation. A folder named internal, a dot directory, or an ignore rule does not establish privacy. An explicit requested deliverable still needs an appropriate audience and home.

Maintain the collection:

  • Organize by reader need and existing conventions: learning, doing, reference, explanation, and consequential decisions may need different forms. Keep the README useful for orientation and navigation. Do not impose a universal folder tree, language, filename style, or document length over project requirements.
  • Verify current claims against their source owners. Dates, TODO markers, file extensions, and missing links are investigation signals, not deletion criteria. Classify affected material as keep, update, merge, move, or retire using content, unique value, consumers, and retention obligations.
  • Consolidate unique information into its canonical owner and link to it. Prefer generated references for mechanically derived facts when the project supports them; retain human explanation that adds value. Do not delete generated outputs required by a published contract, build, or distribution. Every generated document has one source and a check that regenerates it and fails on a diff; never edit the generated copy.
  • Preserve significant decision rationale and future-useful migration or evaluation evidence. Mark superseded decisions honestly rather than rewriting history. Completed plans need retention only when their remaining evidence or obligations justify it. Promote useful conclusions before retiring obsolete task records; do not move clutter into an archive merely to hide it.

Close out the docs:

  1. Update links, indexes, build references, and affected consumers with moves or removals. Preserve redirects only for a real published-link compatibility need; otherwise remove the old path instead of leaving duplicate documents.
  2. Run repository link checks and documentation builds when affected, regenerate generated docs and confirm no diff, run the doc examples' check, verify changed claims against code or tests, and run git diff --check. A tidy tree is not proof that documentation is accurate.

Use task-closeout for the rest of closing a task: scratch files, private material, handoffs and branches.

Load a reference only for its task:

  • Reviewing, writing, or trimming project-owned agent instructions or skills from repository evidence: instruction-review.md.
  • Placing, moving or retiring the pages, examples or agent files of a library, SDK or CLI docs set: developer-docs-site.md.

Use technical-writing for prose and page content (its developer-docs reference for library and CLI docs), research for external evidence, and repository-text-hygiene for Git, encoding, and line-ending drift. This procedure grants no publication, history-rewrite, or deletion authority beyond the user's task.

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.