Templates

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

Organizes and persists project documentation into the appropriate docs/ categories using existing conversation and repository context. Use when the user asks to record work, sync docs, update design/research/implementation/reference/user/context docs, maintain postmortems, archive superseded material, or refresh documentation indexes.

TDD workflow and testing strategy for code implementation. Use when writing tests, implementing features that need tests, or fixing bugs. Defines Red-Green-Refactor gates, integration-first strategy, and test coverage requirements.

调用 codex 命令行进行代码审核，自动收集当前文件修改和任务状态一并发送；工作区干净时自动审核最新提交 (user)

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, optimize a skill's description for better triggering accuracy, or need help with SKILL.md files, frontmatter fields, skill structure, naming conventions, or skill development best practices.

Detailed usage guides for MCP services (Sequential Thinking, Context7, Async MCP, GitHub, Google Developer Knowledge, Codebase Retrieval) and multi-tool collaboration patterns. Use when you need guidance on how to use a specific MCP service or combine multiple MCP services for complex tasks.

Version Bump

This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code.

Multi-AI Review Skill - Uses Gemini CLI and Codex CLI for comprehensive code/paper review, cross-validation, and multi-perspective analysis. Activates when users request review, validation, or analysis using multiple AI models.

发布 GitHub Release，从 CHANGELOG 生成发布公告并更新 Draft Release (project)

Use the JetBrains IDE MCP Server (IntelliJ IDEA 2025.2+) to let an external client drive IDE-backed actions: run Run Configurations, execute commands in the IDE terminal, read/create/edit project files, search via IDE indexes (text/regex), retrieve code inspections for a file, fetch symbol info, perform rename refactoring, list modules/dependencies/repos, open files in the editor, and reformat code. Use when you want IDE-grade indexing/refactoring/inspection instead of raw shell scripting.

Uses the packaged async-agent-backend binary to run Claude, Codex, or Gemini tasks asynchronously or synchronously from this repository. Use whenever the user asks to submit background agent jobs, run headless provider tasks, wait for completion, stream logs, cancel tasks, or expose the local backend over MCP.

Complete Plan/Code workflow for moderate and complex tasks. Use when task complexity is moderate or complex (cross-module design, multi-file changes, architectural decisions). Not needed for trivial tasks.

Pre-implementation quality gates for non-trivial features. Use when implementing moderate or complex features to ensure proper boundaries, minimal structure, and no unnecessary abstraction. Includes exception protocol for skipping gates.

Converts design screenshots and frontend code (HTML/CSS/React) into production Android XML layouts with functional PRD. Supports both new page creation (0-to-1) and extending existing pages (1-to-2). Use when the user provides design mockups, Figma exports, or asks to create/modify Android XML layouts from design specs.

量化分析项目技术债务并通过多角色辩论评估改进优先级，生成健康评分和路线图。当用户要求分析技术债务、评估项目健康度、讨论架构权衡、多视角评审代码质量、或需要从不同专业角色角度辩论技术方案时触发。

Deep structured analysis and planning with script-enforced quality gates. Use when the user explicitly asks for deep planning, thorough analysis before a complex decision, or invokes /deep-plan. Also use when a task involves cross-service architecture, migration strategy, or multi-stakeholder tradeoffs that need rigorous reasoning before implementation. Not needed for routine Plan/Code workflow — that is handled by plan-code-workflow.

Implements Android features following project-specific MVI architecture, custom base classes (BaseBindingFragment/Activity), LoadSir state pages, BRVAH v3.x adapters, and internal utility APIs (onClick, dp, VLog, TranslateResource, collectWhenStarted). Use when developing any Android feature, creating new screens, implementing lists with pagination, or adding TabLayout+ViewPager2 patterns.