Skip to content
Back to search

claude-code-docs

SkillWorks with Claude CodeGitHub

Everything Anthropic ships for builders, archived and updated 4× daily. 3,900+ files. 12 sources. API, Claude Code, MCP, docs, cookbooks, skills, plugins.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/thevibeworks/claude-code-docs with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/thevibeworks/claude-code-docs'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/thevibeworks--claude-code-docs/badge.svg)](/listings/thevibeworks--claude-code-docs)

HTML

<a href="/listings/thevibeworks--claude-code-docs"><img src="/api/listings/thevibeworks--claude-code-docs/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

git clone https://github.com/thevibeworks/claude-code-docs
cd claude-code-docs

Then ask Claude Code anything:

claude "how does the agent loop work?"
claude "show me the MCP spec for tool annotations"
claude "what did the 'Building Effective Agents' post recommend?"
claude "how do I set up hooks in the Agent SDK?"

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 225 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Claude Code Docs
  2. > Auto-updating archive of Anthropic's builder documentation: every source
  3. > that publishes markdown, fetched 4x daily. 4,100 files from six live sources,
  4. > Anthropic's legal and policy pages with their version history, and a frozen
  5. > archive of the engineering blog.
  6. [![fetch](https://github.com/thevibeworks/claude-code-docs/actions/workflows/fetch-claude-docs.yml/badge.svg)](https://github.com/thevibeworks/claude-code-docs/actions/workflows/fetch-claude-docs.yml)
  7. [![review](https://github.com/thevibeworks/claude-code-docs/actions/workflows/claude-review.yml/badge.svg)](https://github.com/thevibeworks/claude-code-docs/actions/workflows/claude-review.yml)
  8. [![license](https://img.shields.io/github/license/thevibeworks/claude-code-docs)](LICENSE)
  9. [![files](https://img.shields.io/badge/files-4%2C100-blue)](#content)
  10. Clone this repo and point Claude Code at it. Every doc, tutorial, cookbook,
  11. skill, and engineering post Anthropic publishes on a markdown surface --
  12. searchable, version-controlled, and offline.
  13. ## Install
  14. ```bash
  15. git clone https://github.com/thevibeworks/claude-code-docs
  16. cd claude-code-docs
  17. ```
  18. Then ask Claude Code anything:
  19. ```bash
  20. claude "how does the agent loop work?"
  21. claude "show me the MCP spec for tool annotations"
  22. claude "what did the 'Building Effective Agents' post recommend?"
  23. claude "how do I set up hooks in the Agent SDK?"
  24. ```
  25. ## Content
  26. Counts are files on disk as of 2026-09-09; `uv run scripts/fetcher.py --tree`
  27. prints them live.
  28. | Source | Section | Files | What |
  29. |--------|---------|------:|------|
  30. | code.claude.com | `--section claude-code` | 198 | Claude Code + Agent SDK docs |
  31. | platform.claude.com | `--section api` | 2,228 | API reference, build guides |
  32. | claude.com/docs | `--section products` | 226 | Claude Tag, Cowork, office agents, connectors |
  33. | modelcontextprotocol.io | `--section mcp` | 347 | MCP spec, SDKs, governance |
  34. | github.com/anthropics | `--section github` | 764 | Cookbooks, skills, plugins, courses, SDK docs |
  35. | support.claude.com | `--section support` | 372 | Help articles |
  36. | anthropic.com/legal | `--section legal` | 92 | Terms, Usage Policy, privacy, Supported Regions; every listed version |
  37. | anthropic.com | frozen 2026-07-08 | 158 | Engineering, research, news posts -- see below |
  38. The legal pages need no markdown variant. Each page embeds its CMS document and
  39. version list in the page data, and Anthropic serves every past version at
  40. `/legal/archive/<id>`. `content/legal/<slug>.md` is the current text;
  41. `content/legal/<slug>/<date>.md` is each version, by effective date. Read the
  42. version list as Anthropic's account, not a record: see
  43. [`content/legal/README.md`](content/legal/README.md).
  44. The anthropic.com blog has no `--section` flag: it is HTML-only, the jina.ai proxy the
  45. fetcher used was removed in July 2026, and `content/blog/` has been a static
  46. archive since 2026-07-08. What is in it, against what anthropic.com's sitemap
  47. lists today:
  48. | `content/blog/` | Archived | Upstream | Coverage |
  49. |---|---:|---:|---:|
  50. | `engineering/` | 25 | 25 | complete |
  51. | `product/` | 4 | 4 | complete |
  52. | `research/` | 72 | 155 | 46% |
  53. | `news/` | 57 | 260 | 22% |
  54. The engineering posts -- "Building Effective Agents", context engineering, tool
  55. use -- are all here. Research and news are not, and will not grow until
  56. something converts HTML to markdown.
  57. ```
  58. content/
  59. en/docs/claude-code/ Claude Code + Agent SDK
  60. en/api/ API reference (1,500+ endpoints)
  61. en/build-with-claude/ Platform features
  62. en/agents-and-tools/ Tool use, agent skills
  63. en/manage-claude/ Admin, billing, managed agents
  64. claude/ Product docs (Claude Tag, Cowork, office agents)
  65. mcp/ MCP protocol spec + community
  66. legal/ Terms, policies, Supported Regions + past versions
  67. blog/ frozen 2026-07-08, not refreshed
  68. engineering/ Building Effective Agents, context engineering, ...
  69. research/ Research papers
  70. news/ Model releases
  71. product/ Product announcements
  72. github/
  73. cookbooks/ 164 recipes + notebooks
  74. skills/ 90 official Agent Skills
  75. plugins-official/ 266 plugin docs
  76. courses/ 80 prompt engineering notebooks
  77. quickstarts/ Deployable app starters
  78. code-action/ GitHub Actions for Claude Code
  79. sdk-python/ Python SDK reference
  80. sdk-typescript/ TypeScript SDK reference
  81. support/ 365 help articles
  82. ```
  83. ## Fetching
  84. Auto-updates four times daily via GitHub Actions. To fetch manually:
  85. ```bash
  86. # Requires: uv (https://docs.astral.sh/uv/)
  87. uv run scripts/fetcher.py # Fetch everything (~3 min)
  88. uv run scripts/fetcher.py --section mcp # MCP spec only
  89. uv run scripts/fetcher.py --section github # GitHub repos only
  90. uv run scripts/fetcher.py --incremental # Skip existing files
  91. uv run scripts/fetcher.py --tree # Show all sources + counts
  92. uv run scripts/fetcher.py --discover # Probe domains for new sources
  93. ```
  94. GitHub repo fetching needs `GITHUB_TOKEN` or `GH_TOKEN` in the environment.
  95. Every fetched source serves a `.md` variant of each page, so nothing is
  96. converted from HTML. That is also why `content/blog/` is frozen: anthropic.com
  97. is HTML-only and the jina.ai proxy path it used was removed in July 2026.
  98. The one exception is `content/legal/`, rendered from the CMS data the legal
  99. pages embed (see above).
  100. See [`sources.json`](sources.json) for the complete machine-readable source
  101. registry.
  102. ## Source Discovery
  103. The fetcher doesn't just download from hardcoded URLs. It probes every known
  104. Anthropic domain for `robots.txt`, `sitemap.xml`, `llms.txt`, and — the
  105. question that decides everything — whether the domain serves `.md` variants at
  106. all. It also enumerates `github.com/anthropics` and watches the `Location`
  107. header on every redirect it follows.
  108. Known domains: `anthropic.com`, `platform.claude.com`, `code.claude.com`,
  109. `support.claude.com`, `modelcontextprotocol.io`, `claude.ai`, `claude.com`,
  110. `academy.claude.com`
  111. **The result is a file, not a log line.** Every full run rewrites
  112. [`discovery.json`](discovery.json) with what exists upstream that
  113. `sources.json` does not. This matters because printing it did not work: the
  114. pipeline had been logging `support.claude.com -> academy.claude.com` four
  115. times a day for weeks, into an Actions log nobody opens, and those 725 pages
  116. were eventually found by a human chasing a dead support article. A discovery
  117. that isn't a diff doesn't reach anyone.
  118. So a new domain, a new `anthropics` repo, or a domain that starts serving
  119. markdown now shows up as a tracked change, gets classified as high-signal, and
  120. opens a PR — the same path a new doc takes. `discovery.json.review` is the
  121. actionable list: reachable, serves markdown, nothing fetches it. Empty is
  122. healthy. Adding a source stays a human decision.
  123. ```bash
  124. uv run scripts/fetcher.py --discover # manual probe; writes the same file
  125. ```
  126. **Sitemaps are treated as incomplete, not authoritative.** Upstream de-indexes
  127. pages it still serves: in July 2026 platform.claude.com dropped every
  128. per-language SDK reference page from both its sitemap and its `llms.txt` while
  129. continuing to edit them, and the archive quietly stopped refreshing 1,560 files
  130. for seven weeks. So every full run also refetches what is already on disk, and
  131. pages that really died are removed by the reaper below rather than by absence
  132. from an index.
  133. ### Reaping
  134. A page removed upstream used to live here forever — the fetcher only ever added
  135. or overwrote. Full runs now delete archived files whose URL returns 404/410 or
  136. the site's HTML shell, with two guardrails:
  137. - **Only markup is deleted automatically.** A file holding real markdown whose
  138. URL has died is content Anthropic removed and we may hold the only copy; it is
  139. reported for a human instead of destroyed by a job that merges its own PRs.
  140. - **A mass-deletion circuit breaker.** More than 200 pages vanishing at once
  141. means an upstream outage, not 200 real deletions — nothing is deleted and the
  142. run fails loudly.
  143. `--no-reap` reports what would go without touching anything.
  144. ### Tombstones
  145. `tombstones.json` records every URL confirmed gone upstream, with the date and
  146. reason. It exists so that a page which died once does not report as a fresh
  147. failure on every subsequent run — 123 standing failures would pin the success
  148. rate at 96.9% and bury the one new breakage that matters. Later runs count
  149. known deaths quietly and print only what changed: pages newly gone, and pages
  150. that came *back* (whose tombstone is then removed automatically). The success
  151. rate is computed over live docs, so it means something.
  152. ## Automation
  153. Two GitHub Actions workflows power this repo:
  154. **[fetch-claude-docs.yml](.github/workflows/fetch-claude-docs.yml)** --
  155. Scheduled every 6 hours. Runs the fetcher, then hands the diff to
  156. Claude Code (via [claude-code-action](https://github.com/anthropics/claude-code-action))
  157. which decides: ignore noise, commit minor fixes directly, or create a PR
  158. for meaningful changes. Sends push notifications via
  159. [barkme](https://github.com/nickchou/barkme-mcp-server) for PRs.
  160. **[claude-review.yml](.github/workflows/claude-review.yml)** --
  161. Triggered on PRs and `@claude` mentions. Reviews changes, merges routine
  162. updates, creates tracking issues for version bumps, and alerts humans only
  163. for breaking changes.
  164. ## Contributing
  165. PRs welcome. The fetcher is a single Python file (`scripts/fetcher.py`)
  166. with no framework dependencies beyond `aiohttp` and `aiofiles`.
  167. To add a new source:
  168. 1. Add the source definition to `sources.json`
  169. 2. Add the fetch logic to `scripts/fetcher.py`
  170. 3. Run `uv run scripts/fetcher.py --section <name>` to test
  171. 4. Update `--section` choices in the CLI
  172. ## Disclaimer
  173. Unofficial mirror for educational and development purposes. Documentation
  174. content is sourced from Anthropic's public sites. For official docs, visit
  175. [code.claude.com](https://code.claude.com) and
  176. [platform.claude.com](https://platform.claude.com). Repository code from
  177. [anthropics/claude-code](https://github.com/anthropics/claude-code).
  178. Redistribution should comply with Anthropic's
  179. [terms](https://www.anthropic.com/legal/commercial-terms).
  180. ## License
  181. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.