发现 编程开发 相关 Skills，比较来源仓库和用途，查看适合你工作流的技能说明。
Use when building multi-day features, avoiding long-lived branches, features taking >1 day, changes touching multiple systems, or high-risk changes needing gradual rollout - break features into deployable incre...
Use when implementing any feature or bugfix, adding tests, fixing flaky tests, refactoring, or changing behavior. Default approach for new features, bug fixes. Exceptions only for throwaway prototypes or genera...
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before as...
Use when reviewing code quality, before refactoring, when complexity increases, when code feels "hard to change", during code review, or when onboarding to unfamiliar code - systematic identification of code sm...
Use when performing 3+ similar operations like file edits, searches, or git operations. Triggers on repetitive tasks, "do the same for", loop-like patterns. Enables parallel tool calls and batch-editor delegati...
Create native Claude Code hooks in ~/.claude/hooks/. Use when adding event-triggered automation (PreToolUse, PostToolUse, Stop, etc.).
Use when encountering any bug, test failure, or unexpected behavior (including race conditions, deadlocks, concurrency issues) - four-phase framework (root cause investigation, pattern analysis, hypothesis test...
Use when starting feature work that needs isolation, or when completing work and deciding how to integrate - covers full worktree lifecycle from setup through merge/PR/discard
Create custom Task subagents in ~/.claude/agents/. Use when defining new agents for Task tool delegation.
Debug memory leaks, profile memory usage, optimize allocations. Use when heap grows unexpectedly, OOM errors occur, or profiling shows memory bottleneck. Covers C++ (Valgrind, ASAN, RAII), Python (tracemalloc,...
Create slash commands for user-initiated workflows. Use when building repeatable /name commands for Claude Code.
Use when executing implementation plans with independent tasks in the current session or facing 3+ independent issues that can be investigated without shared state or dependencies - dispatches fresh subagent fo...
Browser automation via Puppeteer MCP for JS-rendered content
Route tasks to optimal LLM provider (Gemini, Codex, Copilot, Claude)
Create skills that extend Claude with specialized knowledge and workflows. Use when Claude needs to learn a methodology or approach.
On-demand MCP server access via CLI without adding to settings.json. Use when exploring MCP server capabilities before installing, making one-off calls, testing MCP servers during development, or avoiding conte...
Comprehensive PDF manipulation toolkit for extracting text/tables, creating PDFs, merging/splitting, and handling forms.
Control interactive CLI tools that require a real terminal via tmux send-keys. Use for git rebase -i, git add -p, vim, nano, Python/Node REPLs, or any command that hangs waiting for terminal input.
Store universal coding patterns and insights into file-based memory. Use after completing difficult tasks with broadly-applicable lessons that work across ANY project. Use when user says "--coder-store" or "--l...
Retrieve project-specific insights from file-based memory. Use when working on features, encountering domain-specific questions, or user says "--project-recall" or "--recall" (you decide which scope, may use bo...
Retrieve universal coding patterns from file-based memory. Use when starting complex tasks, encountering unfamiliar problems, or user says "--coder-recall" or "--recall" (you decide which scope, may use both)....
Retrieve universal coding patterns from vector database. Use when starting complex tasks, encountering unfamiliar problems, or user says "--coder-recall" or "--recall" (you decide which scope, may use both). Sk...
Store project-specific insights and context into file-based memory. Use after completing tasks that reveal important details about THIS codebase. Use when user says "--project-store" or "--learn" (you decide wh...
Retrieve universal coding patterns from vector database using true two-stage retrieval. Auto-invokes before complex tasks or when user says "--recall". Searches relevant role collections based on task context.
默认按已收录数据中的 GitHub 仓库 Stars 排序。Stars 属于来源仓库，同一仓库的多个技能可能共享相同数值；它反映仓库关注度，不代表单个技能的安装量或质量评分。