TsukiNi22/ai-utils / skills/docs/docs-update
docs-update
UnverifiedBring every piece of documentation and distribution text of a project up to date with the changes made (and the latest version) - README, CHANGELOG, docs/ HTML site and project graph, gh-pages branch, GitHub wiki, packages / manifests / version numbers, setup / install scripts, --help and usage texts, completions, man pages, examples, templates, marketplace / plugin lists, CI docs, license year - by finding every place the changes touch, not only the obvious ones. Use whenever the user asks to update / refresh / synchronize the docs, "mets à jour toute la doc / le README / le changelog / le site / le wiki / les packages / le script d'install", after a feature, a fix, a release or a version bump.
Update all the documentation
Goal: after changes, nothing that describes the project is stale - docs, site, wiki, packages, install scripts, version strings. Think of every place the project is described, not only the README.
1. Baseline: what changed
Find what the docs must catch up with, in this order:
- What the user says changed (the request, the conversation).
- The pending diff (
\git status,\git diff HEAD) and the commits since the last docs update or the last release:\git describe --tags --abbrev=0,\git log <tag>..HEAD --oneline,\git log -1 --format=%H -- README.md CHANGELOG.md docs. - The latest version: the source of truth is the code (
project(... VERSION x.y.z),pyproject.toml,package.json,Cargo.toml, thereference/VERSION.mdof a skill (libutils), the last tag, aversionconstant). Take the newest one found, and report any disagreement between them. For a dependency documented from its upstream (ex: libutils), take its latest release (libutilsskill:scripts/update.sh).
Run scripts/inventory.sh [repo] for the first pass: it lists the documentation surfaces that exist, the
versions found, dead links in the Markdown and what is not committed yet.
2. Inventory every surface that exists
Check each one; skip what the project has not. Do not stop at the first hit.
| Surface | What to bring up to date | Skill |
|---|---|---|
README.md (+ sub-READMEs of tools / modules) | features, tables, command lists, install, usage, layout tree, badges, versions, links | readme-style |
CHANGELOG.md | new entry for the changes / version (its own commit) | git-conventions |
| Release notes, tag text | the next release body (not published without a request) | git-conventions |
docs/ HTML (user guide, technical doc, project graph) | pages, summaries, numbered cards, graph data regenerated, EN/FR texts, theme | html-doc (+ html-style) |
gh-pages branch / site / packages repository | the published copy of docs, graph, repo metadata, index of packages (worktree or clone, see below) | html-doc, cpp-project |
GitHub wiki (<repo>.wiki.git) | Home version stamp, sidebar, section pages, member tables | wiki-style |
| Packages and manifests | version in CMakeLists.txt, CPack / spec / control, pyproject.toml, package.json, Cargo.toml, plugin / marketplace JSON (names, descriptions, skill lists), Dockerfile labels | cmake-style, python-project |
| Setup / install / update scripts | usage header, command and option lists, list of installable items, URLs, default versions, uninstall paths | coding-style |
--help / usage text, man pages, shell completions | every option and command that exists now | the style skill of the language |
Examples, templates, reference/ files, skill and router tables | outdated names, removed options, new items | the skill's own rules |
| CI / CD docs and workflow comments | job names, keywords, images, versions | ci-cd |
LICENSE / notices | year, holder, third-party notices after a dependency change | license, audit-deps |
| Doxygen / docstrings / code comments describing changed behavior | the stale ones only | comments |
| claude.ai bundles, generated archives, exported files | rebuilt from the sources (never edited by hand) | the project's build script |
3. Update
- For each change, ask "where else is this written?":
grep -rnthe old name, option, path, version and count (ex: "9 routers", a list of skills) over the whole repository, the wiki andgh-pages. - Check every name / command / option / API against the code before writing it; mark planned vs implemented.
- Same list written in several places (README table, layout tree, plugin table, router tables, marketplace JSON, setup script header, claude.ai README): change them all in one pass so they stay identical.
- Generated files are regenerated with their generator (graph, archives, reference from a commit), not edited.
- Follow the style skill of each surface; texts in English unless the project's docs are not.
gh-pagesand the wiki are other branches / repositories: usegit worktree addor a temporary clone, edit there, then remove the worktree / clone and come back on the original branch; leave the result uncommitted unless the user asked for commits (thengit-commitfor each repository).- No version bump, tag, release or publication on your own: if the version in the docs must move, say which files and to what, and do it only when asked.
4. Verify
- Re-run
scripts/inventory.sh: no dead link, versions identical everywhere. - Anchors and tables (
wiki-stylehascheck_wiki.py;html-stylehascheck_style.py, which also checks the responsive rules of the HTML). - The commands written in the docs actually run (
--help, install in a scratch prefix when cheap). - Scripts / builds that read the docs' lists still work (ex: rebuild the archives,
./setup.sh list).
5. Report
List by surface what was changed (path, one line), what was checked and found already current, what could not be
done (private wiki, no network, version decision) and the commit command to suggest (git-commit). A written
summary of a wide pass goes through report only when the user wants a file.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization. | Stars: 464k | Updated | Claude CodeCodex | Verified | Install |
| academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). | Stars: 196k | Updated | Claude CodeCodex | Unverified | Install |
| doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |