Branch and commit conventions that keep an agent's work reviewable — one branch per task, main stays clean, nothing force-pushed.
中文简介正在准备中。当前显示原文说明。
Look up how Claude Code itself works — hooks, skills, subagents, MCP, plugins, settings — from the official docs, so answers come from the current spec instead of memory.
中文简介正在准备中。当前显示原文说明。
查看安装方法安装前可以查看交给 AI 代理的指令内容。
Answer questions about Claude Code from the official documentation, fetched at the moment you need it. Do not answer from memory: this product ships changes weekly, and a confidently wrong answer about a hook event or a manifest field costs more than a fetch.
Fetch the page that covers the question with WebFetch. Start from the overview only if
you need to find your way around; otherwise go straight to the relevant page.
| Topic | URL |
|---|---|
| Overview / index | https://code.claude.com/docs/en/overview |
| Skills | https://code.claude.com/docs/en/skills |
| Hooks — guide | https://code.claude.com/docs/en/hooks-guide |
| Hooks — full reference | https://code.claude.com/docs/en/hooks |
| Subagents | https://code.claude.com/docs/en/sub-agents |
| MCP | https://code.claude.com/docs/en/mcp |
| Plugins — guide | https://code.claude.com/docs/en/plugins |
| Plugins — reference | https://code.claude.com/docs/en/plugins-reference |
| Plugin marketplaces | https://code.claude.com/docs/en/plugin-marketplaces |
| Slash commands | https://code.claude.com/docs/en/slash-commands |
| Settings | https://code.claude.com/docs/en/settings |
| CLI reference | https://code.claude.com/docs/en/cli-reference |
If a URL 404s, the docs have been reorganised — fetch the overview and follow its navigation rather than guessing a new path.
Verify against the live CLI where you can. For anything about plugins, hooks or settings, the installed binary is a second source of truth and it cannot be out of date:
claude plugin validate <path> # what the schema actually accepts, warnings included
claude plugin --help # current subcommands
claude --help # flags
When the docs and the CLI disagree, trust the CLI and say so in your answer.
WebFetch needs network access. With none, say that you are answering from memory and
that the spec should be re-checked — do not present a remembered field list as current.
If you want a local copy for offline work, scripts/scrape_claude_code_docs.py in this
repo mirrors the docs into docs/claude-code/ for personal use. That mirror is
deliberately not committed or shipped: it is Anthropic's content, and a stale local copy
is exactly how wrong answers get confident.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Branch and commit conventions that keep an agent's work reviewable — one branch per task, main stays clean, nothing force-pushed.
中文简介正在准备中。当前显示原文说明。
Design better agents, skills and hooks for Claude Code, checking the current spec instead of guessing at fields.
中文简介正在准备中。当前显示原文说明。
Run a feature end to end — scout the code, plan it, build it, and keep a session file so a closed terminal doesn't lose the thread.
中文简介正在准备中。当前显示原文说明。
Close out a session — summarise what changed, capture what was learned, and prepare the commit.
中文简介正在准备中。当前显示原文说明。
Hand over a spec and walk away — plans, builds, tests and audits on its own, stopping only at a real blocker.
中文简介正在准备中。当前显示原文说明。
Run the project's own tests, linter, type checker and build, and report what actually failed.
中文简介正在准备中。当前显示原文说明。