Skills 导航

Bootstrap a multi-agent software project from idea to running CI/CD. Use when starting a new project that needs agent team design, task management, GitHub repo setup, TDD pipeline, and Discord notifications. Triggers on "new project", "bootstrap project", "set up agents for project", "create project pipeline", "start a new repo with CI/CD".

Use when the user wants to pressure-test a product, internal-tool, or OSS concept against Amazon's Working Backwards PRFAQ gauntlet before committing to a spec. Triggers on: \"work backwards\", \"write a PRFAQ\", \"press release first\", \"is this idea worth building\", \"pressure-test this concept\", \"filter this before brainstorm\", \"is this a real product\". Also catches solution-first pitches (\"I want to build X that does Y\") and technology-first pitches (\"use AI to...\") that need customer-first filtering. Produces a binary pass/fail verdict, not a polished doc. Hardcore coaching — direct, skeptical, concrete. On pass, hands off to brainstorm-beagle with a concept brief. Does NOT write code, plan implementation, scaffold projects, or draft specs.

Deep first-principles analysis of any topic, decision, strategy, or assumption. Strips inherited thinking, identifies what is provably true, and rebuilds from ground truth. Use when user asks for first principles analysis, wants to challenge assumptions, says "analyze this from scratch", "break this down", "what's really true here", or triggers with /firstprinciples. Also useful for strategic decisions, investment theses, product strategy, career moves, or any situation where conventional wisdom may be wrong.

Analysis of Bot Protection systems with available countermeasures 🚿. How to defeat anti-bot system 👻 browser fingerprinting, javascript, abck, arkose, automation, aws-waf-token, bot. Use when you need browser fingerprinting capabilities. Triggers on: browser fingerprinting.

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.

This skill should be used when the user wants to compare and search for flight or train ticket prices across multiple platforms. It supports both domestic (China) and international routes, fetches real-time train availability via 12306, generates direct search links for all major booking platforms and airline official websites, provides WeChat mini program quick links for mobile search, and highlights discount conditions. For flight prices, it attempts web scraping first (no API key needed), then falls back to Tequila/Amadeus API for existing key holders. Trigger scenarios include: searching for cheap flights, comparing train ticket prices, finding international flight deals, looking for the best ticket booking platform, or asking about ticket discount conditions.

Analyze and price Canadian residential properties using comps, price-per-square-foot reasoning, market context, and pricing strategy. Use when estimating home value, setting a list price, comparing comparable properties, evaluating sale scenarios, or advising sellers, buyers, or investors in Canada.

Adversarial verification of AI-generated work. Spawns an independent verifier to check for false claims, broken code, and security issues.

Fetch review comments from a PR and evaluate with receive-feedback skill

多 Agent 项目管理 - 从想法到立项，自动在 Discord 频道创建项目 Thread 并分配 Agent 任务。支持快速讨论路由、批量立项、项目状态追踪。Triggers: '立项', '创建项目', '新项目', '讨论一下', '项目状态', 'create project', 'discuss', 'project status'

营销文案助手。标题公式(100个)、正文写作、AIDA框架、痛点挖掘、客户证言、落地页文案。Copywriting with 100 headline formulas, body copy, AIDA framework, pain points, testimonials.

并行调度 Claude Code agent teams 与 Codex CLI，从 UX、API、架构、文档等多视角审查产品，汇总成可执行优化方案，适合产品自查、发布前审查和信息架构审计。

✎ 做产品审查或发布前把关时很省心，多代理并行从体验、信息架构到竞品多视角分析，最后沉淀成可执行优化方案。

适合把模糊需求、零散 prompt 提炼成可测试的结构化规范，基于 EARS 补全触发条件、约束和示例，用于优化 AI 生成代码、产品方案或文档输入。

✎ 把含糊需求快速整理成清晰、可执行的规范，尤其适合写代码和产品文档时用 EARS 方法补齐细节、少走弯路。

专门整理臃肿的 CLAUDE.md，用渐进式披露把高频规则留在主文档、细节移到 references，建立多入口索引和校验流程，适合给 Claude Code 的项目记忆做降噪和长期维护。

✎ 把复杂提示词、文档和工作流按步骤渐进展开，能显著缓解上下文拥挤，让 AI 协作更稳，也更适合长期维护。

用Promptfoo搭建和执行LLM评测流程，覆盖promptfooconfig.yaml配置、Prompt测试、模型对比、Python自定义断言、llm-rubric评分与few-shot示例管理。

✎ 做提示词测试和模型对比时，提示词评测能把 Promptfoo 流程快速搭起来，YAML 配置、Python 断言和 LLM-as-judge 都很顺手。

扫描代码库中的 DRY 重复、KISS/YAGNI 失衡、错误处理缺口和 DI 设计问题，输出带严重级别、位置、修复成本、模式签名与重构建议的审计报告和评分。

✎ 想系统揪出代码里违反 DRY、KISS、YAGNI、错误处理和依赖注入的问题，用它能按严重级别、位置与修复成本给出审计结果，排查特别高效。

支持从空目录创建新项目，或把现有原型重构成生产可用结构，自动补齐 Clean Architecture、Docker、CI/CD、质量与安全配置。

✎ 不管是从零起步还是改造老项目，项目脚手架都能快速搭出可上线的生产级结构，省下大量规范化和 DevOps 配置时间。

结合MCP Ref、Context7和WebSearch快速查最佳实践、官方文档与版本信息，并生成guide、manual、ADR或research文档，适合给项目补齐模式说明、API手册和架构决策记录。

聚焦生产级NestJS开发规范，覆盖模块边界、依赖注入作用域、异常过滤器、DTO校验与Drizzle ORM集成，适合架构设计、重构评审和团队统一代码风格。

聚焦高级 Prompt 设计与优化，覆盖 few-shot、chain-of-thought、模板体系和 system prompt，适合为复杂推理任务构建可复用、可评估、可上线的生产级提示方案。