行動開發

Intelligent skill router that analyzes user requests and automatically dispatches to the most appropriate skill(s) or zen-mcp tools. Routes to zen-chat for Q&A, zen-thinkdeep for deep problem investigation, codex-code-reviewer for code quality, simple-gemini for standard docs/tests, deep-gemini for deep analysis, or plan-down for planning. Use this skill proactively to interpret all user requests and determine the optimal execution path.

Collaborative documentation and test code writing workflow using zen mcp's clink to launch gemini CLI session in WSL (via 'gemini' command) where all writing operations are executed. Use this skill when the user requests "use gemini to write test files", "use gemini to write documentation", "generate related test files", "generate an explanatory document", or similar document/test writing tasks. The gemini CLI session acts as the specialist writer, working with the main Claude model for context gathering, outline approval, and final review. For test code, codex CLI (also launched via clink) validates quality after gemini completes writing.

Complete App Store Optimization (ASO) toolkit for researching, optimizing, and tracking mobile app performance on Apple App Store and Google Play Store

Ensures questions are answered literally before taking action. Triggers on user input containing '?' or patterns like 'why did you...?', 'will that work?', 'have you considered...?'. Use when user asks about your decisions, challenges an approach, or requests assessment. Prevents interpreting questions as implicit instructions or criticism.

Strict test-driven development state machine with red-green-refactor cycles. Enforces test-first development, meaningful failures, minimum implementations, and full verification. Activates when user requests: 'use a TDD approach', 'start TDD', 'test-drive this'.

Upgrade dependencies for Java/Kotlin (Gradle/Maven) and TypeScript/Node projects with minimal risk: plan the bump, apply changes incrementally, run tests/builds, and document breaking changes. Use when the user asks to bump deps, update frameworks, or address CVEs.

Plan work before coding: do repo research, analyze options/risks, and ask clarifying questions before proposing an implementation plan. Use when the user asks for a plan, design/approach, scope breakdown, or implementation steps.

Systematic approach for gathering authoritative, version-accurate documentation. Claude invokes this skill when research is needed before implementation. Ensures truth over speed while achieving both.

Manages practice rules. Use when user states a preference or approach, or asks to add/modify rules for coding, architecture, tooling, or best practices.

Prevents overengineering through worthiness scoring, opportunity cost comparison, and branch threshold monitoring. Triggers: scope creep, overengineering, worthiness score, branch size, YAGNI, feature evaluation, scope validation, anti-overengineering, opportunity cost Use when: evaluating features during brainstorming, planning new functionality, branches approach size limits (1000/1500/2000 lines, 15/25/30 commits) DO NOT use when: feature is already approved and in progress. DO NOT use when: simple bug fixes with clear scope. Use this skill BEFORE implementing any new feature. This is NON-NEGOTIABLE for scope control.

Reduce context usage with MECW principles (keep under 50% of total window). Triggers: context pressure, token usage, MECW, context window, optimization, decomposition, workflow splitting, context management, token optimization Use when: context usage approaches 50% of window, tasks need decomposition, complex multi-step operations planned, context pressure is high DO NOT use when: simple single-step tasks with low context usage. DO NOT use when: already using mcp-code-execution for tool chains. Use this skill BEFORE starting complex tasks. Check context levels proactively.

Process external resources into actionable knowledge with evaluation, storage, and application decisions. Triggers: knowledge intake, article evaluation, paper review, external resource Use when: user shares links to articles, papers, or external resources DO NOT use when: searching existing knowledge - use knowledge-locator.

Structured file enumeration and content analysis for understanding codebase structure before reviews or refactoring. Triggers: file analysis, codebase structure, directory mapping, hotspot detection, code exploration, file enumeration, structure mapping, module boundaries Use when: before architecture reviews to understand file organization, exploring unfamiliar codebases to map structure, estimating scope for refactoring or migration DO NOT use when: general code exploration - use the Explore agent. DO NOT use when: searching for specific patterns - use Grep directly. Provides structural context for downstream review and refactoring workflows.

Fix GitHub issues using subagent-driven-development with parallel execution where appropriate. Triggers: fix issue, github issue, issue resolution, subagent development, parallel execution, issue workflow, gh issue, fix github Use when: addressing GitHub issues systematically, multiple related issues need fixing, tasks can be parallelized across subagents, quality gates needed between task batches DO NOT use when: single simple fix - just implement directly. DO NOT use when: issue needs clarification - comment first to clarify scope. Uses subagents for parallel execution with code review gates between batches.

Architecture-aware project initialization combining online research with archetype selection. Triggers: project initialization, new project, architecture decision, project setup, which architecture, choose architecture, project architecture Use when: initializing a new project and need to select an appropriate architecture based on project type, team size, domain complexity, and current best practices DO NOT use when: architecture is already decided - use project-init instead. DO NOT use when: exploring multiple architectures - use architecture-paradigms instead.

Orchestrate and run appropriate pensive review skills based on codebase analysis and context. Triggers: code review, unified review, full review, review orchestration, multi-domain review, intelligent review, auto-detect review Use when: general review needed without knowing which specific skill applies, full multi-domain review desired, integrated reporting needed DO NOT use when: specific review type known - use bug-review, test-review, etc. DO NOT use when: architecture-only focus - use architecture-review. Use this skill when orchestrating multiple review types.

Use when user asks to leverage claude or claude code to do something (e.g. implement a feature design or review codes, etc). Provides non-interactive automation mode for hands-off task execution without approval prompts.

Write self-documenting code with clear structure and naming, adding minimal, helpful, evergreen comments only for large sections of complex logic, avoiding comments about recent changes or temporary fixes. Use this skill when writing or reviewing code in any file where comments or code documentation are involved. Apply when deciding whether to add comments, when refactoring code to be more self-explanatory through better naming and structure, when reviewing existing comments for relevance and evergreen quality, when removing outdated or change-related comments, or when adding concise explanatory comments for complex logic sections. Use for any task involving code documentation strategy, comment quality assessment, or improving code readability through structure rather than comments.

Expert knowledge for deploying, managing, and troubleshooting AWS Elastic Beanstalk applications with production best practices

Use when creating or fixing .claude/rules/ files - provides correct paths frontmatter (not globs), glob patterns, and avoids Cursor-specific fields like alwaysApply