菜单
Use when answering questions about current Claude Code behavior, configuration, CLI features, hooks, skills, plugins, sessions, compaction, permissions, or other Claude Code-specific interfaces from official documentation.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Jeremy's coding style and conventions (fastai style guide + 3 decades of experience), used across ALL his projects. TRIGGER — read before the first tool call that reads, writes, or searches files: the patterns are needed to discuss and assess code that's read or found, not only to write it. Tasks that never touch the filesystem don't need it. Certainly BEFORE writing or editing any Python: a new file, a one-line change, a test, a refactor, anything. Don't skip because a change "looks trivial" or you're "just" fixing one line. Covers naming, layout, concision, imports, testing process (TDD red-green, pytest), docments, versioning, and the fastcore/fasthtml ecosystem preference.
Workflow and gotchas for nbdev projects, where notebooks in nbs/ are the source of truth. TRIGGER — read BEFORE touching code, tests, or docs in ANY nbdev repo (has a [tool.nbdev] section in pyproject.toml): before the first cell edit, not after. Don't edit the autogenerated .py files (projects often also mix in plain hand-written .py modules - check the per-file marker). Routes to the pyskill docs that carry the substance, and holds the nbdev v3 delta and environment specifics. Assumes and pulls in the persistent-python and coding-patterns skills.
The persistent clikernel Python session — Jeremy's preferred single tool for nearly ALL Python work. TRIGGER — read and bootstrap as step zero of EVERY session where clikernel is available, before the first tool call: even tasks that only read a file or search a directory, plus stateful inspection, editing code or notebooks, debugging, experiments, API probes, data transforms, pyskills discovery, notebook-style work, local tool orchestration. If it could go through Python, it goes here — never python -c, Bash heredocs, or ad hoc scripts.
How to write prose that doesn't read as AI slop. TRIGGER — read BEFORE writing or editing ANY prose meant for a human to read: READMEs, docs, articles, PR/commit text, emails, announcements, release notes, or comments longer than a line. Not for code. Covers a marked good/sloppy sample pair, the numbered tells they illustrate, banned words, and the em-dash and hard-wrap bans.
Use when answering questions about current Codex behavior, configuration, CLI features, hooks, skills, plugins, sessions, compaction, permissions, or other Codex-specific interfaces from official documentation.
Jeremy's coding style and conventions (fastai style guide + 3 decades of experience), used across ALL his projects. TRIGGER — read before the first tool call that reads, writes, or searches files: the patterns are needed to discuss and assess code that's read or found, not only to write it. Tasks that never touch the filesystem don't need it. Certainly BEFORE writing or editing any Python: a new file, a one-line change, a test, a refactor, anything. Don't skip because a change "looks trivial" or you're "just" fixing one line. Covers naming, layout, concision, imports, testing process (TDD red-green, pytest), docments, versioning, and the fastcore/fasthtml ecosystem preference.
基于 SOC 职业分类
| name | claude-code-docs |
| description | Use when answering questions about current Claude Code behavior, configuration, CLI features, hooks, skills, plugins, sessions, compaction, permissions, or other Claude Code-specific interfaces from official documentation. |
clikernel. Fetch https://code.claude.com/docs/llms.txt exactly with httpx, save its text to a variable, and display the whole variable. It is the table of contents agents are meant to inspect.httpx and save its text to another variable. Never fetch llms-full.txt; select the relevant page from llms.txt instead.len(). If it is at most 30,000 characters, display the whole page. If it is larger, do not display it: load the toolslm.read_md pyskill, parse the page variable, inspect paths(), and retrieve the relevant sections. Fetch shared sections separately when an event-specific section refers to them.--help, /doctor).