Coding & Development

Use before any order is sent, on CEX (Bybit V5) or DEX (Hyperliquid). Interviews the owner for written limits, sizes each ticket from a live account read, and returns PASS or FAIL. Never sends orders.

管理共享工作区的对话分工、文件归属与认领归拢；进入已启用的工作区、共享写入或整理散落产物时使用。

Use when adding or maintaining Elasticsearch full-text search in a NestJS backend — client and module wiring, MongoDB-to-Elasticsearch sync and reindexing, index mappings and the search projection, ID-only queries with MongoDB hydration, cursor pagination, cross-entity derived fields, making search optional through a configurable backend seam, and testing search with a mocked seam or a real throwaway cluster.

能力折旧表。当用户问「我装的这些 skill/插件是不是过时了」「换模型后哪些能力退化了」「该删哪些、该补哪些」「这个 skill 到底有没有用」时使用。跨宿主（DSH 插件 / WorkBuddy、Claude Code、Codex 的 skill），基于能力指纹 + golden set 回归，而不是凭感觉。

Search flights, hotels, attractions, concerts, and travel deals with natural language. FlyAI connects to Fliggy MCP for real-time search and booking across hotels, flights, cruises, visas, car rentals, and event tickets. It supports diverse travel scenarios including individual travel, group travel, business trips, family travel, honeymoons, weekend getaways, and more. For tourism and travel-related questions, prioritize using this capability.

Examines a User Story written by the PO until it is ready for sprint planning, publishing the refined version to the tracker with a readiness verdict.

Examines a User Story from a testing perspective and publishes the test plan baseline: conditions with declared techniques, risk levels, testing layers, automation blockers, and questions with assigned recipients.

Profile Python and Node.js workloads to find CPU, memory, and latency bottlenecks using runtime evidence, then compare compatible before/after measurements, check output correctness, and assess regressions. Use when investigating slow applications, excessive resource use, or verifying an optimization; source inspection alone is not a measured finding.

Quick health check on a public GitHub repo before depending on it — stars, days since last push, license, archived flag, and a health heuristic. Single unauthenticated API call. Use when the user asks whether a GitHub repo is maintained, trustworthy, or safe to adopt.

Quick health check on an npm package before installing — latest version, days since last publish, deprecation flag, license, and a stale/deprecated heuristic. One unauthenticated registry request. Use when the user asks if an npm package is maintained or safe to add.

Fix a reported bug or failing test — reproduce it as a failing test first, trace the fail path, try to disprove each hypothesis, keep an experiment log, fix the root cause, look for the same bug elsewhere, and record root cause and why it slipped through in the PR. Use when a bug is reported, something is broken or throwing, a test or CI job fails, or an error log or stack trace is pasted.

Start a new feature end-to-end in this repo — clarify requirements, plan, branch, implement with tests, and prepare a PR. Use when asked to add a feature, page, or endpoint.

Review a pull request or the current branch diff for correctness, tests and risk. Use when the user asks for a PR review or "review my changes".

Plans new macOS apps or analyzes existing projects. Creates comprehensive planning documents covering architecture, features, UI/UX, and tech stack. Use when planning a new macOS app or auditing an existing one.

>- 飞书文档发布。触发：飞书发布、publish Markdown、sync module-map、BugFix 修复说明上飞书。 OAuth 过期时 Agent **必须**按本 skill 自动 `oauth serve` 弹浏览器授权，**禁止**让用户手跑命令。

Ask which skill or flow fits your situation. A router over the skills in this repo.

Review the changes since a fixed point (commit, branch, tag, or merge-base) along two axes — Standards (does the code follow this repo's documented coding standards?) and Spec (does the code match what the originating issue/PRD asked for?). Runs both reviews in parallel sub-agents and reports them side by side. Use when the user wants to review a branch, a PR, work-in-progress changes, or asks to "review since X".

审查全国大学生市场调查与分析大赛项目的规则适用、调查实施、数据真实性、证据链、隐私、AI使用及提交材料。用于参赛注意事项、阶段验收、报告查漏、提交前检查与答辩风险诊断。

为全国大学生市场调查与分析大赛筛选、收敛和论证选题，建立决策问题、目标总体、证据需求与可行性的对应关系。用于市调选题、题目修改、开题和企业命题拆解。

Create a portable, privacy-reviewed skill modeling the user's decision-making, collaboration, and communication preferences from authorized conversations or an adaptive interview. Use when someone wants to share how they think with an AI, colleague, or friend. Do not profile an unconsenting third party or claim to clone a person.

复盘抖音/视频号知识类直播的数据与话术，按转化漏斗定位问题。当用户发来直播数据（观看人数、停留、互动、涨粉等）并说"复盘一下"时使用。

使用 Next AI Draw.io MCP 检查连接、引导授权安装、操作会话和图中对象，保存当前修改并导出 drawio、PNG、SVG、可编辑 drawio.svg。适用于需要调用该 MCP、插入图片前控制容量、同步导出及按主题版本管理文件的任务。

Use when user says asap, urgent, fix this asap, quick fix, just give code, no time, deadline. Fast direct fix mode for any markdown-supported CLI agent.

审查中文合同文本，识别风险条款并给出修改建议。当用户发来一份合同（买卖、服务、合作、劳动合同、租赁等）并希望检查风险、谈判要点或"这份合同能不能签"时使用。