lennney

"AI 审查 / AI code review. 用户说"帮我审查代码"、"review 一下"、"看看这段有没有问题"时加载。在后台检查 bug、安全、性能、代码风格，然后告诉用户结果。"

"AI 编码 / AI coding. 用户说"帮我写个组件"、"实现这个功能"、"写个 API 路由"时加载。先问清需求，在后台实现，完成后自动调 /ai-code-review 审查。"

"AI 调试 / AI debugging. 用户说"帮我 debug"、"报错了"、"这个 bug 怎么修"、"抛异常了"时加载。让用户描述问题，你在后台分析根因并修复。"

"AI Git 操作 / AI git workflow. 用户说"帮我 commit"、"提个 PR"、"push 一下"、"git 操作"时加载。在后台生成 commit message / PR 描述，用户确认后执行。"

"AI 重构 / AI refactoring. 用户说"重构这段代码"、"优化代码结构"、"把这块抽出来"时加载。先让用户定范围，你在后台改完自动验证。建议先跑 /tdd 保证测试覆盖。"

"AI 测试 / AI testing. 用户说"加测试"、"写单元测试"、"测一下这个函数"时加载。在后台分析代码、写测试、跑通。对核心逻辑用 TDD 风格。"

Profesional web & landing page builder with Next.js 16, React 19, Tailwind CSS v4, shadcn/ui, and Framer Motion. 用 Next.js 16 + React 19 + Tailwind CSS v4 + shadcn/ui 构建专业落地页。8 种视觉方向、10 套色板、10 组字体搭配。 用户说"帮我建个落地页"、"做一个官网"、"设计一个 landing page"时加载。

Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, wants to be grilled, or says "is this ready", "review my plan", "有什么漏洞", "帮我审查一下这个方案".

AI 审查。用户要审查代码时加载。你在后台检查完告诉用户结果。

AI 编码。用户要写代码时加载。让用户描述要做什么，你在后台实现。

AI 调试。用户 debug 时加载。让用户描述问题，你在后台分析修复。

AI Git 操作。用户需要 commit/PR 时加载。你在后台生成好，用户确认即可。

AI 重构。用户要重构代码时加载。先定范围，你在后台改完验证。

AI 测试。用户要加测试时加载。你在后台分析代码、写测试、跑通。

Create frontend UIs that don't look AI-generated. Use when building new pages or reviewing design output for "AI taste."

Build accessible, responsive, and performant frontend components with design system best practices, modern CSS, and framework-agnostic patterns.

Use when auditing TicketPilot project plans, modules, tests, OpenSpec changes, and phase status before continuing development.

Use after every implementation batch to produce a strict acceptance report with requirement-to-evidence mapping.

Use after every meaningful feature, bug fix, architecture decision, evaluation update, prompt change, or skill change.

Use when modifying retrieval, chunking, RRF, rerank, citation, answer grounding, or evaluation scripts.

Use when adding API providers, environment variables, external services, logs, deployment configs, or documentation that may expose secrets.

Use when planning or implementing any non-trivial TicketPilot change. Requires proposal, design, tasks, acceptance criteria, and validation before implementation.

Use when implementing or reviewing risk assessment, human review routing, complaint handling, compensation handling, legal risk, privacy risk, or account security risk.

Use when deciding project scope, prioritization, or whether a feature belongs in the TicketPilot MVP.

Keep coding agents focused on requested and necessary work. Use for bounded changes, review-only tasks, scope creep, speculative hardening, unnecessary hashing or dependencies, repeated audit loops, or when the user invokes Stop That Shit.

Use Agent Search MCP for free-first English and Chinese web search with compact evidence and minimal tool calls. Trigger for factual discovery, claim verification, Chinese web results, selected-page extraction, provider failures, freshness limits, or search…

Use when reviewing product proposals, technical designs, SEO content strategies, or any decision needing multiple expert perspectives. Multi-agent cross-functional review with SEO-specific roles (E-E-A-T, Google policy, keyword strategy). Independent scoring…

Post-task reflection and knowledge management system. Auto-extracts errors and solutions, builds searchable knowledge base.