About claude-code-docs
claude-code-docs is a skill backed by a local mirror of Claude Code documentation. It is for product and configuration questions, with Markdown pages refreshed by a GitHub Action every three hours and retrieved by topic.
Key features
- 01Local Markdown documentation mirror
- 02Refreshes every three hours
- 03Topic index and targeted page lookup
- 04Covers settings, commands, and extensions
- 05Source links and stale-content fallback
Use cases
- 01Look up Claude Code CLI behavior and settings
- 02Find guidance on MCP, hooks, skills, or plugins
- 03Check release notes and documentation sources
MCP support
The skill includes documentation topics about MCP in Claude Code. It is not itself an MCP server.
Install
Copy the command and run it in a terminal.
npx skills add https:/ / github.com/ mehmetbaykar/ claude-code-docs-skill --skill claude-code-docsAfter installing, type/claude-code-docsin Claude Code to confirm it loaded.
Details
From the project READMEClaude Code Docs
Local mirror of Anthropic Claude Code documentation, kept fresh by a 3-hour GitHub Action. The cleaned Markdown lives in references/; the auto-generated topic list lives in references/INDEX.md; the per-file manifest with upstream URLs lives in references/docs_manifest.json.
Scope
Use this skill for Claude Code-specific product and configuration questions, including CLI behavior, settings, slash commands, hooks, skills, plugins, MCP, subagents, AGENTS.md, output styles, sandboxing, permissions, environment variables, monitoring and usage, Claude Desktop, Slack, Chrome, Web, Agent SDK, and release notes. If the question is about OpenAI Codex, Cursor, Claude.ai consumer features, the general Anthropic API, or another non-Claude-Code product, this skill does not apply.
Workflow
- If the user supplied a topic, normalize it to a slug:
- lowercase, strip leading
/docs/en/, strip surrounding slashes - join nested segments with
__(e.g.agent sdk hooks->agent-sdk__hooks,whats new 2026 w19->whats-new__2026-w19)
- lowercase, strip leading
- If
references/<slug>.mdexists, read that file directly. Do NOT grep the wholereferences/tree first - the index plus targeted reads is faster and uses less context. - If no exact match, read
references/INDEX.mdand pick the closest topic. If still ambiguous, list the candidates and ask. - If the user supplied no topic, read
references/INDEX.mdand present the available topics.
Answer format
- Lead with a direct answer to the user's question grounded in the file you read.
- Quote short snippets (commands, config keys) when they appear verbatim in the doc.
- End with
Source: <upstream URL>using theoriginal_urlfrom the file frontmatter orreferences/docs_manifest.json.
Freshness and fallback
The mirror is refreshed every 3 hours by upstream CI, which fails rather than committing frozen content. If the local content looks stale, contradicted by the user, or empty:
- Suggest the user run
npx skills update claude-code-docs. - Check the file's entry in
references/docs_manifest.json: astatusofstalemeans upstream could not be reached on the last run. - Cross-check the canonical URL via
original_urlinreferences/docs_manifest.jsonand offer it as a follow-up source. - If a specific page failed MDX cleaning, the unmodified source is preserved at
references/_raw/<slug>.md-- read that as a fallback.
Examples
| Invocation | Reads |
|---|---|
$claude-code-docs hooks / /claude-code-docs hooks | references/hooks.md |
$claude-code-docs mcp | references/mcp.md |
$claude-code-docs agent sdk hooks | references/agent-sdk__hooks.md |
$claude-code-docs cli reference | references/cli-reference.md |
$claude-code-docs changelog | references/changelog.md |
$claude-code-docs (no argument) | references/INDEX.md |