agentleFS
CsHengCsHeng/agent-skills/skills/organize-docs/SKILL.md

organize-docs

Skill3 starsChanged 8 days agoMITView on GitHub

Use for docs organization: README/AGENTS ownership, one-way legacy CLAUDE.md cleanup into AGENTS.md, stable truth roots, docs layout, docs/.ignore, stage artifacts, canonical terminology, search boundaries, and clear Markdown and diagrams written in their intended form. Recording already verified behavior without restructuring belongs to sync-truth.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 9 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    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:

    default docs search needs a local search-boundary policy such as docs/.ignore

    line 14
  2. Very specific:

    Read-only project-state explanation and local git or worktree status go to analyze-project.

    line 19
  3. Distinctive:

    Keep stable truth roots and stage artifact roots explicit.

    line 26
  4. Distinctive:

    Keep default docs search free of stage artifacts when the repository needs that search boundary.

    line 27
  5. Distinctive:

    Treat stage artifacts as history: keep them while a decision, operation, or reference needs them.

    line 28
  6. Distinctive:

    Keep live design and plan prose to goals, authorized discretion, ownership, delivery endpoints, and pause conditions, and cite superseded …

    line 29
  7. Distinctive:

    When an authorized change creates, promotes, supersedes, compacts, or retires durable decision truth, read Decision Record Lifecycle and …

    line 30
  8. Distinctive:

    When consolidating plan artifacts, inspect the source roots in scope, keep useful material under the repository's canonical stage layout, …

    line 31
  9. Distinctive:

    When a repository has competing names for one concept, define the canonical term in stable docs, and update docs, paths, tests, and code …

    line 32
  10. Distinctive:

    When authoring or revising Markdown prose or diagrams, read Writing Stable Docs.

    line 34
  11. Distinctive:

    Write the requested Markdown in its final form following Writing Stable Docs.

    line 40

≈ “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

Organize Docs

Maintain a concise account of project intent, decisions, and current behavior within the requested documentation work or the documentation scope of an authorized owning workflow. Edit docs only under the request or approved scope; drift findings feed that decision. Apply development-standards Maintainability for how code behavior and document intent relate.

Use This Skill When

  • the user wants to reorganize README.md, AGENTS.md, existing legacy CLAUDE.md, or stable docs, or change what each document owns
  • the repository needs explicit stable truth roots and stage artifact roots
  • default docs search needs a local search-boundary policy such as docs/.ignore
  • scattered plan, draft, or execution-note roots are to be consolidated into one stage-artifact tree, when requested or part of the authorized cleanup
  • stable docs, paths, tests, or code need canonical terminology alignment
  • user-approved drift follow-up from analyze-project points to stable doc maintenance

Read-only project-state explanation and local git or worktree status go to analyze-project; recording already verified behavior without changing ownership or structure goes to sync-truth.

Rules

  • Treat a task as docs governance when it changes document ownership, truth roots, layout, search boundaries, or terminology, and keep cleanup to the authorized touch set. Mutate on explicit user intent, a user-approved drift follow-up, or the owning workflow's existing documentation authority (including sync-truth handoffs).
  • README.md stays human-facing; AGENTS.md is the maintained AI-facing truth root; CLAUDE.md is a legacy input, not a truth root.
  • Record the repository root CLAUDE.md path type (absent, regular file, symlink, unsafe type) before mutation. When the requested scope covers an existing root CLAUDE.md (an explicit legacy-cleanup request, a user-approved drift follow-up naming that path, or an authorized owning workflow's documentation scope), follow Legacy CLAUDE.md Migration; the final state is a valid AGENTS.md and no root CLAUDE.md path, including a dangling symlink. Otherwise maintain AGENTS.md only and report an uncovered legacy file.
  • Keep stable truth roots and stage artifact roots explicit. When repository policy names an external owner for designs, plans, evaluations, or archives, resolve its declared root; keep product-required guidance self-contained so standalone product checks do not need the external checkout.
  • Keep default docs search free of stage artifacts when the repository needs that search boundary.
  • Treat stage artifacts as history: keep them while a decision, operation, or reference needs them; otherwise delete them during authorized cleanup and repair affected links.
  • Keep live design and plan prose to goals, authorized discretion, ownership, delivery endpoints, and pause conditions, and cite superseded gates shortly from stage history. Migrate historical plans only when cleanup is the request.
  • When an authorized change creates, promotes, supersedes, compacts, or retires durable decision truth, read Decision Record Lifecycle and classify current status together with future value.
  • When consolidating plan artifacts, inspect the source roots in scope, keep useful material under the repository's canonical stage layout, and repair references after moves or deletions; an explicitly retained historical original may stay unchanged.
  • When a repository has competing names for one concept, define the canonical term in stable docs, and update docs, paths, tests, and code references together.
  • Name intentionally retained historical or reference material archived, and compatibility surfaces targeting an older, alternate, or constrained version compat.
  • When authoring or revising Markdown prose or diagrams, read Writing Stable Docs.

Workflow

  1. Assess the current layout: README.md, AGENTS.md, the recorded root CLAUDE.md path type, docs/, and local docs policy files such as docs/.ignore; classify stable truth roots versus stage artifact roots using repository-local policy first.
  2. Apply the rules above to the requested work, moving or summarizing content into stable docs domains with plans, drafts, and other stage artifacts kept as history.
  3. Write the requested Markdown in its final form following Writing Stable Docs.
  4. Validate as below.

Validation

Run repository-owned checks for the boundaries affected by the change; review prose and diagram meaning against current code, observed behavior, and the intended change, since link and path checks cover links, ownership, and placement. When documentation truth boundaries change, run the bundled boundary checker; when inspecting a declared documentation manifest, use the advisory layout checker; for an authorized mechanical wrapping repair of existing prose, use the bundled normalizer. Read Documentation Checks And Repair for those invocations. For a few affected lines, edit directly; repair only observed drift within the authorized scope and leave preserved history unchanged.

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.