Claude Skills by BlueSkyXN

为 OpenAI Codex 设计、审计或重构仓库的 AGENTS.md 文件。当用户要求 '写 AGENTS.md'、'配置 AGENTS.md'、'为 Codex 准备项目指令'、'AGENTS 分层'、'按 Codex 加载模型整理仓库指令'、'审计 AGENTS.md' 时使用。前提是团队默认从仓库根目录启动 Codex。不适用于：CLAUDE.md、Cursor 规则、其他 agent 配置格式；也不要在用户只是问 'AGENTS.md 是什么' 时触发——只在要求写、改、审计时触发。

Use when converting data analysis results into a clear report with assumptions, findings, validation, limitations, and next steps.

Use when planning safe data cleaning, deduplication, missing-value handling, anomaly handling, and reproducible cleaned outputs.

Use when reviewing or creating SQL for analysis, metric definitions, joins, filters, aggregation, and validation queries.

Use when analyzing CSV, Excel, parquet, or table-like files and producing reproducible summaries.

Use for accessibility review of web or app UI, including semantic structure, keyboard navigation, focus management, ARIA, labels, contrast, responsive behavior, and WCAG-oriented validation.

Use for designing or reviewing API contracts, including REST, GraphQL, OpenAPI, auth, pagination, versioning, errors, and backward compatibility.

Use for software bug fixes that need reproduction, root-cause analysis, minimal implementation, and validation.

Use for diagnosing or improving build speed, bundle size, cache behavior, monorepo task execution, packaging output, and reproducible builds.

Use for designing, implementing, reviewing, or testing CLI tools, command surfaces, flags, config precedence, terminal output, and exit behavior.

Use for dependency audits, vulnerability remediation, version conflicts, package manager migrations, lockfile updates, and safe upgrade plans.

Use for implementing or reviewing frontend UI, components, routing, state, responsive layout, accessibility, browser behavior, and API integration.

Use for planning and implementing features that span frontend, backend, API, database, scripts, or tests.

Use for branch strategy, isolated worktrees, commits, PR preparation, merge/rebase decisions, conflict handling, tags, and release-oriented Git hygiene.

Use for legacy modernization, framework upgrades, incremental rewrites, service extraction, schema migrations, and risky technical migration planning.

Use for diagnosing slow code, APIs, queries, memory growth, frontend load, build regressions, resource bottlenecks, and scalability limits.

Use for reviewing a branch, pull request, local diff, or implementation plan before merge.

Use for designing, testing, comparing, versioning, and validating prompts or LLM app behavior with measurable criteria, datasets, cost tracking, and safety constraints.

Use for Python implementation, modernization, typing, async, packaging, pytest, ruff/mypy checks, scripts, and Python performance or maintainability work.

Use for planning software refactors before editing files, especially when behavior must be preserved.

Use for pre-release validation, versioning, changelog/release notes, tags, build artifacts, smoke tests, rollback notes, and release readiness review.

Use for first-pass software repository onboarding: map structure, entry points, tech stack, commands, tests, risks, and the smallest safe next actions before editing.

Use for software security review of code, configs, workflows, auth, permissions, input validation, secrets, dependency risk, and supply-chain surfaces.

Use for designing, improving, or validating software test plans, unit/integration/e2e coverage, automation, CI checks, flaky tests, and regression safety.

Use when producing concise decision-ready briefing notes from processed research materials.

Use when converting meeting notes, transcripts, or chat logs into minutes, decisions, action items, and open questions.

Use when turning materials into a presentation storyline, slide outline, speaker notes, or sharing-session structure.

Use when drafting project status reports, milestone reports, risk summaries, or management updates from provided materials.

Use when drafting or reviewing weekly reports from provided materials, progress logs, meeting notes, or team updates.

Use to convert approved PRD/spec/tech/task artifacts into repository change-spec assets such as proposal, design, tasks, and spec deltas.

Use to turn approved specs or tech briefs into task cards, dependency graph, test tasks, and agent handoff. Not for code.

Use for behavior specs from a PRD or scope lock: flows, states, fields, permissions, UI states, errors, ACs, and traceability.

Use for idea-to-PRD work: concept PRD, scope lock, detailed PRD, and spec handoff. Not for API, schema, tasks, or code.

用于读取已有代码项目，结合真实代码、文档、local 材料和 git 记录，产出项目功能能力地图：项目定位、一级能力域、二级功能组或模块、三级功能点、证据路径、当前状态、下一步，以及可选的完全汉化 XLSX 阅读版。默认不做工作分解、产品结构拆解、成熟度评估、卡点长分析、任务计划、排期或资源估算。

Use to check whether PRD, spec, tech bridge, task plan, or change-spec artifacts are ready for the next phase. Reviews only.

Use to turn functional specs into engineering bridge artifacts: architecture context, API/data drafts, NFRs, test draft, and open questions.

Use when building a claim/evidence/source table from provided research materials.

Use when turning mapped evidence and provided materials into a balanced research synthesis.

Use when deduplicating overlapping research materials, AI notes, excerpts, transcripts, and source packs.