my-claude-code-settings

Plan-first brainstorming workflow that turns an idea into an approved implementation plan before coding. Use when the user wants to brainstorm, design, scope, stress-test, or plan a feature/spec. Spark adapts to the active planning surface: in Codex native Plan mode it stays read-only and returns a chat-only final plan; in Claude Code Plan mode it uses Claude's plan approval/exit flow and writes files only after approval exits Plan mode; in writable/default mode it saves Markdown plans under .plannings/YYYY-MM-DD-feature-slug.md. HTML or visual artifacts are explicit and writable-mode-only.

Write, debug, and validate ast-grep structural code search rules. Use this skill when the user needs syntax-aware code search, AST pattern matching, structural refactor discovery, language-construct queries, or searches that plain text tools like rg can miss, such as finding functions with particular descendants, calls inside specific contexts, missing error handling, React hook shapes, decorators, or other Tree-sitter-backed code structures.

Audit and improve Codex AGENTS.md guidance files and companion code_map.md navigation maps in repositories. Use when the user asks to check, audit, update, optimize, or fix AGENTS.md files; asks for Codex project guidance maintenance; mentions nested AGENTS.md conflicts, code maps, stale commands, scoped instructions, sandbox or approval boundaries; or says "优化 AGENTS.md", "审计 AGENTS.md", "检查 nested AGENTS.md", "更新 Codex 项目指导", or "生成 code_map.md".

Archive root-level planning files (`task_plan.md`, `findings.md`, and `progress.md`) into a timestamped `.plannings/` directory for the current feature. Use when Codex should run `$archive-planning [feature-name]`, close out an active plan, preserve completed planning context, or replace the deprecated Codex prompt workflow with a deterministic skill script.

Audit and improve Claude Code CLAUDE.md guidance files, .claude/rules/ path-scoped rules, and companion code_map.md navigation maps. Use when the user asks to check, audit, update, optimize, or fix CLAUDE.md files; mentions nested CLAUDE.md, additive loading, ancestor/descendant loading, path-scoped rules, @import chains, CLAUDE.local.md, claudeMdExcludes, or code_map.md; or says "优化 CLAUDE.md", "审计 CLAUDE.md", "检查嵌套 CLAUDE.md", "Claude Code 项目指导", "生成 code_map (Claude)". Make sure to use this skill whenever the user wants to manage Claude Code memory files at any layer, even if they only mention CLAUDE.md without explicit audit wording.

Analyze a repository and current Codex environment, then recommend Codex CLI, Codex App, AGENTS.md, skills, native subagents, plugins, MCP servers, config/hooks, and optional OMX workflow improvements without modifying files. Use when the user asks to optimize Codex workflows, improve Codex setup, recommend Codex automation, configure MCP/plugins/subagents, analyze AGENTS.md scope needs, or says "优化 Codex 流程", "Codex 工作流推荐", "给 Codex 配 MCP", "配置 Codex plugins", or "配置 Codex subagents".

Create self-contained HTML artifacts, single-file by default and split into offline per-page bundles when oversized, for complex, visual, comparison-heavy, reviewable, or shareable work outputs. Use this skill when the user asks for an HTML artifact, browser-viewable report, implementation plan, PR/code review report, architecture explainer, design comparison, research briefing, incident report, status dashboard, temporary structured-data editor, mini deck, or design-system specimen. Prefer this skill for long Markdown-like outputs that benefit from navigation, cards, diagrams, filters, copy/export buttons, or annotated diffs. Do not use it for short answers, simple commands, commit messages, tiny patch summaries, or production UI implementation unless the user explicitly asks for a review artifact.

Modify the Beautiful Mermaid live editor itself rather than writing ordinary Mermaid diagrams. Use when the task mentions the Beautiful Mermaid repo, `editor.ts`, generated `editor.html`, config panel/options, themes or dark mode, zoom, PNG/SVG export, clipboard behavior, sample presets, or renderer wiring for the editor. Do not use for generic Mermaid syntax help or normal Markdown Mermaid authoring.

Author Typst slide decks with Touying. Use when working with .typ presentations, themes, or animations.

Address GitHub PR review comments and actionable review threads with GitHub CLI. Use when triaging reviewer feedback, summarizing unresolved PR comments, or applying selected fixes on a pull request.

Initialize GitHub repository configuration from vetted upstream templates. Use when setting up repository automation, issue and PR templates, CI workflows, or baseline GitHub project files for a new or existing repo.

Debug and fix failing GitHub PR checks with GitHub CLI. Use when GitHub Actions checks fail, when a PR has broken status checks, or when you need local reproduction commands for CI failures.

Safely orchestrate Conventional Commits for staged Git changes, or for all working-tree changes when the user explicitly asks to include everything. Use when the user asks to write a commit message, split staged changes, split all changes, commit everything regardless of stage state, include untracked files in the commit set, organize a messy index before committing, or generate structured commit text without pushing by default. Output language is auto-detected from the user's instruction, the request's own language, and the repository's recent commit history (English fallback); explicit phrases like `用中文提交`, `commit in English`, or `请使用中文拆分提交所有的改动` override detection. Agent commits automatically inject `Agent-Task` / `Agent-Model` / `Generated-By` trailers — plus optional `Confidence` / `Scope-risk` / `Tested` quality trailers — and an `[AI]` header tag; Why-line is required for feat/fix/refactor/perf; supports a `chore(wip)` checkpoint mode for long-running agent tasks.

Multi-source deep research skill for current-topic investigation, comparison, and cited report writing. Use when the user asks to research, compare, or deep-dive a topic with current web sources and citations, including phrases like research, deep dive, latest, 调研, 做个深度研究, or 带来源总结. Do not use for offline-only codebase questions, casual opinion requests, or when the user explicitly forbids web access.

Use when the user asks to ground an ambitious proposal, avoid over-grand designs, make a bold direction executable, pressure-test feasibility, prevent "too much vision and too little landing", or turn a strategy/refactor/product idea into the smallest verifiable first move with stop rules. Trigger for requests such as 落地, 先落地, 别太飘, 收一收, 可执行, 可验证, 止损, and for follow-ups after geju-style big-picture thinking. Do not trigger for ordinary code review or implementation unless the user explicitly asks to ground or shrink the plan first.

Use when the user explicitly asks to think bigger, open up the design space, challenge conservative design, avoid over-indexing on backward compatibility, escape local-detail fixation, or make a bold high-level product or architecture direction call. Use for strategic reframing, not for ordinary code review, PRD writing, implementation planning, or adversarial risk review.

Implement safe, behavior-preserving code refactors after inspecting the existing project. Use when the user asks to refactor code, split large files or modules, extract functions or methods, reduce duplicated logic, rename confusing classes/functions/variables, improve code comments, remove unused or dead code, or says 重构代码, 拆分模块, 提取方法, 减少重复代码, 优化命名, 优化注释, 删除未调用代码. For broad refactor requests, plan safe slices and wait for approval; for narrow scoped requests, directly implement the smallest verifiable slice.

Use only when the user explicitly asks for swarm, subagents, parallel agents, dynamic workflow, multi-agent orchestration, 多智能体编排, or when the task truly needs coordinated research plus implementation plus review plus verification packets. Do not use for ordinary code review, planning-only work, single-line bugfixes, routine audits, or migrations unless orchestration is requested or at least two independent workflow dimensions are present.

Run a code quality review focused on maintainability, structure, abstraction quality, file growth, branching complexity, boundary cleanliness, and refactoring opportunities. Use when the user asks for code quality review, code review, maintainability review, architecture quality review, PR code quality feedback, 代码质量审查, 代码质量 review, 可维护性审查, 架构质量审查, or review comments about code structure. Do not use for pure security review, formatting-only review, performance profiling, or implementation tasks unless the user also asks for a code quality review.

Review a codebase for architectural friction, rank deep-module refactoring opportunities, and draft RFCs for safer interfaces and boundary-test strategies. Use when the user wants to improve architecture, identify refactoring seams, consolidate tightly coupled modules, deepen shallow modules, redesign an interface around a core concept, replace brittle unit tests with boundary tests, or turn an architecture review into an RFC or issue draft. Chinese trigger phrases also apply: 架构改进, 抽象太浅, 模块拆得太碎, 接口太多, 测试钉死实现细节, 想写架构 RFC, 这块代码改一改, 模块边界乱.

Structured code review across correctness, security, performance, readability, testing, and architecture, with language-specific guidance and human-readable findings. Use whenever the user asks to review a PR, inspect git changes before merge, audit a directory or file set, prepare merge feedback, summarize review findings, or do code review / PR review / CR / review comments / 代码审查. Adapt the output language to the user's context: use Chinese review wording for Chinese or mixed Chinese discussions, and English review wording for English-first discussions.

Challenge ideas, requirements, technical plans, products, pricing, markets, pitch/BP narratives, and major personal decisions with a no-flattery adversarial review. Use when the user asks for 泼冷水, 挑刺, 骂我, 给我泼盆冷水, 假设你是我的对手, 帮我 challenge 一下, 我这个想法有什么问题, 哪里会崩, 魔鬼代言人, devil's advocate, challenge assumptions, 别夸我, hidden assumptions, failure modes, overengineering, missing evidence, or kill criteria before implementation or commitment. Do not trigger for ordinary code review unless the user explicitly asks to challenge assumptions or failure modes.

Use when the user wants to compact context before auto-compaction kicks in, hand off an unfinished task to a fresh session, switch topics mid-stream and preserve state, or pick up yesterday's work in a new conversation. Produces a single handoff.md that captures completed work, blocked items with what was already tried, next concrete steps, and the minimum context needed to resume cold. Use proactively when context usage is high, a long-running task hits a natural break, or the user signals a session switch. Triggers include 上下文压缩, 交接文档, handoff, 续接会话, 主题切换, 接力, 明早接着干, context handoff, compact before, fresh session, continue tomorrow. Do NOT use for single-file fixes that finish in one commit (put it in the commit body), Q&A or pure investigation sessions, or end-of-session retrospectives meant for human review (those are different artifacts).

Use when implementing a multi-step spec, PRD, design doc, GitHub issue, or approved plan where decisions, deviations, and tradeoffs accumulate during coding. Maintain a live implementation-notes.md alongside the work capturing design decisions (choices made where the spec was ambiguous), intentional deviations from the spec, alternatives considered and rejection criteria, and open questions for human review. Use proactively the moment implementation of a written spec begins, before the first edit. Triggers include 实施记录, 实现笔记, 决策日志, 边写边记, 按这个 spec 实现, 按这个计划做, implementation log, decision log, spec divergence, implement this plan. Do NOT use for single-file fixes, pure investigation, one-off scripts, work without a written spec to deviate from, or simple tasks where one commit captures the reasoning.

智能标书编写专家，专精工程咨询、建筑设计、市政工程领域的投标文件编写。 当用户提到：标书、投标、招标、投标文件、技术标、商务标、招标响应、投标方案、 bid document、bid proposal、tender、proposal writing、编写标书、写标书、 招标文件分析、评分标准优化 时使用此技能。

Write or update technical documentation from the real codebase and project files. Use whenever the user asks for README, API docs, architecture guides, user guides, CONTRIBUTING docs, migration notes, or JSDoc/code comments, and also when they want technical documentation rewritten into natural Chinese with correct terminology and formatting.

Researcher-profile-driven paper intake and literature workbench for academic workflows. Use this whenever the user wants to skim, deep-read, card, compare, synthesize, map research gaps, or build a literature review from papers, arXiv/AlphaXiv links, DOIs, PDFs, landing pages, or existing paper JSON / workbench artifacts. Normalize sources into `paper-record`, then route into scan, deep-read, card, synthesis, review, or compatibility modes (`json`, `interpret`, `xray`). Trigger even when the user only says things like “精读这篇”, “整合这几篇”, “找研究空白”, or “搭综述框架”.

Multi-perspective roundtable discussion skill for exploring a topic through a truth-seeking moderator and a small set of representative real figures. Use when the user asks for 圆桌讨论 / roundtable / 多人物观点碰撞 / 让几位思想家讨论 / 模拟多方思想交锋 / 结构化辩论探索, or wants to examine one topic through disciplined multi-party debate rather than a plain summary. Do not use it for ordinary argumentative writing, generic pros/cons lists, or a single-speaker explanation.