菜单
初始化新项目的 Claude Code 配置。用户说"初始化项目"、"init project"、"配置 Claude Code"时触发。
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。
基于 SOC 职业分类
Deep critical analysis of one specific paper — dissect experimental setup, extract numbers, evaluate claims against a hypothesis. Triggers: 'analyze this paper', 'deep dive into', '分析这篇论文', '论文深度分析', 'what does this paper really show', '这篇论文说明了什么', arXiv ID with analysis intent. Not for paper discovery (use scholar-agent), quick summaries, or literature surveys.
Check whether documentation needs updating after code changes. Triggered when user says "sync docs", "文档同步", or "检查文档".
Manage ML experiment lifecycle via structured YAML registry — register, record results, compare runs, track status. Triggers: 'register experiment', 'compare experiments', 'experiment results', 'exp init', 'exp list', '注册实验', '对比实验', '查实验', '哪个实验最好', '记录结果'. Not for training monitoring, checkpoints, or launching jobs.
Pick the next executable task from project-root TODO.md by priority + dependency, mark it in-progress, then start executing. Explicit invocation only — never auto-run. Triggers: 'work on todo', 'next task', 'pick a todo', '领 todo', 'todo worker', '接下来做什么', '领下一个任务'. NOT for: listing TODO contents (just read it), brainstorming what to do, interrupting active coding.
Self-review for paper paragraphs — diagnose v1 issues and give a revision direction (intro, abstract, method, related work). Triggers: 'review this intro', 'check this paragraph', 'paper paragraph self-review', 'help me look at intro / abstract', 'polish this intro', 'paper self-review', '帮我看 intro', 'paper 段落自检', 'review 这段 paper'. Not for: generating a draft from scratch — write v1 yourself first, then iterate.
Audit project context management (CLAUDE.md, rules, knowledge) for progressive disclosure compliance. Checks that all knowledge files are reachable from rules, detects stale references, orphaned files, and CLAUDE.md index leakage. Triggers: 'audit context', 'check context', 'context hygiene', 'context audit', '检查 context', '审计上下文', 'knowledge 覆盖检查'. NOT for: code review, rule content correctness, or knowledge content quality.
| name | init-project |
| description | 初始化新项目的 Claude Code 配置。用户说"初始化项目"、"init project"、"配置 Claude Code"时触发。 |
Automatically configure Claude Code best practices for new projects: directory skeleton, general-purpose agents, hooks, and AGENTS.md (with CLAUDE.md as a thin @AGENTS.md stub).
Phase 1: Generate Skeleton (deterministic script)
→ Run scripts/init-skeleton.sh to create directory structure + boilerplate files
→ Output which files were created
Phase 2: Interactive AGENTS.md Fill (LLM + AskUserQuestion)
→ Process <!-- init-project: 待填充 --> placeholders section by section
→ For each section: read codebase → generate draft → AskUserQuestion to confirm → write
→ User can reply "skip" to skip any section
→ CLAUDE.md is just a stub (@AGENTS.md + write-routing rule); never edited in Phase 2
→ Detailed guide templates in details/agents-md-sections.md
Phase 3: Profile Overlay (optional)
→ AskUserQuestion: "Want to overlay an additional profile? Currently supported: research"
→ If research selected → run scripts/init-research-profile.sh
→ Profile details in details/research-profile.md
Output Summary → List all generated/modified files → Suggest next steps: review content, git add, start developing
| Section | Auto-exploration | User prompt |
|---|---|---|
| Project overview | Read README, pyproject.toml, package.json | "One-line description of this project's core goal?" |
| Models / Key dependencies | Detect imports, API keys, model configs | "What models or external services does this project depend on?" |
| Directory structure | ls + read key file docstrings | "Does this directory layout look correct? Anything to adjust?" |
| Dev workflow | Detect CI, Makefile, scripts/ | "Use the default brainstorming→plans→dev→verify flow?" |
| Dev guide | Detect venv, .env, Dockerfile | "Any special environment setup steps?" |
| Context Loading (routing table) | List .claude/rules/*.md + scan opening summaries | "Confirm (task domain → rule) mapping — entry point for Codex / external agents" |
| Always Do (project-specific) | Read rules/, lint config | "Any cross-module consistency requirements?" |
| Ask First | Scan core files (interfaces, config) | "Which files/dirs require confirmation before modifying?" |
| Never Do | Detect third_party/, .env | "Any absolute don't-touch conventions?" |
| Progressive disclosure setup | Scan docs/knowledge/ + .claude/rules/ | Ensure rules reference knowledge files inline; ask about missing refs |
<!-- init-project: 待填充 --> placeholders are touchedinit-skeleton.sh and init-research-profile.sh work independently of the skilldetails/skeleton-manifest.mddetails/agents-md-sections.mddetails/agent-templates.mddetails/research-profile.md