powershell-bridge
Use when asked to set up a local PowerShell bridge for controlled command execution on Windows.
prompt-runner
Run NAVIGATION/PROMPTS tasks by parsing headers, enforcing prompt canon gates (lint, FILL_ME__, hash checks, allowlists, dependencies), executing declared commands, and emitting deterministic receipts/reports. Use when asked to execute a prompt file or run a phase prompt.
catalytic-wormhole
Integrated catalytic wormhole compression pipeline using the full lab stack (boundary stress, graph isomorphism, phase cavity sieve, KV cache compression, orthogonal multimodel). Compresses LLM eigenbasis rotation chains by cavity-sieving signal from noise.
pi-harness
Run the local Pi coding agent as a persistent, headless worker with stable session IDs, background tasks, status polling, result collection, cancellation, and follow-up prompts in the same session. Use when an agent or local automation needs to delegate work to Pi, let it continue in the background, check progress later, or resume the same Pi conversation across multiple turns.
arxiv-to-md
Use when downloading, converting, or fetching an arXiv paper as markdown. Handles arxiv IDs and URLs with LaTeX/pandoc or HTML conversion methods.
coderabbit-comments
Retrieve CodeRabbit review comments from the VS Code extension's local storage.
Copy a recipe to an Obsidian vault
Given a URL to a recipe, copy it to an Obsidian vault using these rules.
Find for designs / scopes for a project
Find relevant designs and scopes for a project. Should be used whenever a user mentions a design / scope doc for a project.
Initialize New Python Project
Setup a new python project in this directory.
Write to Obsidian Vault
Use this skill when the user asks to write, create, or save a note, idea, draft, list, or any other content to their Obsidian vault.
plan-implement
Use this skill to implement a plan in full, updating the plan as you go, with continuous testing
plan-iterate
Use this skill when you've added notes to a plan and want them addressed
plan-todos
Use this skill to add a detailed todo list to your plan
Unit Test Writing
This skill should be used when the user asks to "write tests", "add unit tests", "create test cases", "test this function", "add test coverage", or discusses testing strategies and test implementation.
intent-analysis
Analyze and explain the intent behind AI tool calls. Use when you need to understand what an action will do, verify your interpretation is correct, or explain your reasoning to the user.
github-workflow-automation
Automates GitHub workflows including batch issue creation from JSON/CSV, PR creation with issue linking, pre-PR testing, and WordPress-specific validation (PHP linting, WPCS). Use when creating multiple issues, automating PR workflows, or validating WordPress code before commits.
archive-work
Archive completed scratchpads and session logs to project history. Invoke when user says "archive this work", "clean up scratchpad", "archive scratchpad", or after PR is merged.
commit-changes
Create thoughtful git commits with conventional commits format. Invoke when user says "commit", "commit these changes", "create a commit", "save my changes", or after completing a task.
create-issue
Create GitHub issues from natural language, designed for ad hoc idea capture mid-flow. Invoke when user says "create an issue", "file a bug", "open an issue to track", "add a feature request", or "create a ticket".
create-pr
Create context-aware pull requests with issue integration. Invoke when user says "create a PR", "open pull request", "make a PR", "submit for review", or after completing work on a feature branch.
do-work
Execute development work from a scratchpad, tracking progress with TodoWrite and making atomic commits. This skill MUST be invoked as the first action when user says "start work on issue #X", "do work on issue #X", "work on issue #X", "continue work", "resume work", or "keep working". Do not fetch issue details or plan manually.
prime-session
Orient to current project by reading CLAUDE.md and architecture docs. Auto-invokes when Claude detects a new or unfamiliar repository. Can also be triggered by "orient me", "what is this project", or "prime session".
review-pr
Conduct comprehensive PR reviews with sprint/roadmap context awareness. Invoke when user says "review PR", "check this PR", "review pull request #X", or when asked to provide feedback on a PR.
setup-work
Setup development environment from GitHub issue. Invoke when user says "init work on issue #X", "initialize issue #X", "setup issue #X", "prepare issue #X", "start fresh on issue #X", "start new work on issue #X", or provides a GitHub issue URL.
stash-artifact
Save ad hoc scripts, notes, or other artifacts to the project's context directory (outside the code repo). Invoke when user says "stash this", "save this script", "keep this outside the repo", "save to context", "stash artifact", or when user has a script or note they want preserved but not committed.
principle-soc
实现关注点分离，确保不同职责模块化和边界清晰。当需要架构分层设计、模块职责划分、横切关注点处理时使用。覆盖层次分离、功能分离、横切关注点三大维度，提供分离检查清单和耦合问题的诊断修复方法。
principle-kiss
保持设计和实现的简单性，识别并消除不必要的复杂度。当需要架构设计评审、代码简化重构、接口设计优化时使用。覆盖代码、架构、接口、流程四大维度，提供简单性检查清单和过度复杂性的诊断修复方法。
arch-options-compare
比较多个架构方案（单体/微服务等），从SOLID原则、成本、复杂度等维度评分，生成对比表和推荐方案。当需要向客户解释架构选择或ADR决策时使用。
tech-selection-explain
将技术选型决策转换为业务语言解释，生成客户友好的技术说明。当ARCH中技术术语过多、需要向PM/客户解释时使用。
vs-coverage-check
检查每个US是否有对应的VS，生成覆盖率报告和修复建议。当设计文档创建后、CONSTRAINT验收前使用，避免US→VS覆盖率<100%导致验收失败。
tdd-cycle
执行 TDD 红绿重构循环（RED→GREEN→REFACTOR→REPEAT），通过先写失败测试、最小代码实现、保持绿色重构的节奏驱动开发。当启动功能开发、需要测试驱动节奏、进行迭代编码、追求高测试覆盖率时使用。提供检查清单和节奏控制。
goal-granularity-check
确保GOAL粒度合理，不过粗或过细。当GOAL创建完成后使用，确保GOAL符合单一职责原则。
goal-sc-coverage-check
检查VS→GOAL→SC覆盖完整性，确保每个VS都有对应的GOAL和SC。当GOAL和SC创建完成后使用，避免覆盖缺口导致验收失败。
goal-traceability-validator
验证GOAL的source_vs追溯链完整性，确保每个GOAL都能追溯到原始VS。当GOAL创建完成后使用。
acceptance-criteria
检查Acceptance Criteria格式和完整性，验证是否符合Given-When-Then结构、覆盖正常流程/边界条件/异常场景。适合在为User Story编写AC后、准备测试用例前使用，当需要验收AC质量时。帮助不熟悉BDD的PM/BA确保AC明确、可测试、覆盖完整，避免遗漏关键场景。
interview-to-us
将访谈记录、会议纪要、邮件沟通等非结构化内容快速转换为符合CRAFT标准的User Story草稿。适合在需求调研阶段使用，当有大量访谈笔记需要结构化时。帮助不会写US的PM/BA快速上手，从自然语言访谈记录自动提取As-Want-So结构，节省手动整理时间（1小时→10分钟）。
nfr-to-user-impact
将技术性的NFR（非功能需求）约束转换为用户能感知的影响描述，用业务语言解释技术约束的价值。适合在NFR定义完成后、需要向客户或管理层解释时使用，当NFR过于技术化时。帮助不熟悉技术术语的PM/客户/BA理解NFR的业务意义，让技术约束变得亲切易懂。
web-app-thinking
提供网页应用全栈架构思考框架，涵盖前端渲染策略、后端 API 设计、基础设施部署、安全防护、性能优化五大维度。当需要设计完整 Web 应用、评审网页系统架构、或需要全局视角审视前后端协同设计时使用。支持 SPA/MPA、SSR/CSR、REST/GraphQL、容器/Serverless 等多种技术栈决策。
principle-solid
应用面向对象设计五大原则（单一职责、开闭、里氏替换、接口隔离、依赖倒置），系统化评审类与模块设计质量。当需要评审架构设计、重构现有代码、审查类职责划分时使用。支持多层次检查清单和违反原则的诊断修复。
principle-yagni
识别并消除过度设计，确保只实现当前明确需要的功能。当需要评审功能范围、重构冗余代码、权衡技术债务时使用。覆盖功能必要性、过度设计信号、技术债务权衡三大维度，提供具体的识别标准和删减建议。
document-quality
多维度审查 Markdown 文档质量，覆盖结构、内容、格式、引用、元数据五大维度。当需要评审文档质量、统一格式规范、验收交付文档时使用。提供分级检查清单、通过标准、常见问题修复指南，支持从快速检查到深度审计的多层次需求。
principle-dry
识别并消除知识重复，确保系统中每条知识有唯一权威表示。当需要重构代码、整理文档、优化配置、建立单一数据源时使用。覆盖代码重复、数据冗余、文档重复三大维度，提供具体消除策略和权衡建议。
principle-invest
评估用户故事质量的六维度框架（独立性、可协商、有价值、可估算、小粒度、可测试），系统化审查需求质量。当需要编写用户故事、评审需求文档、拆分史诗故事时使用。提供分级检查清单和低质量故事的改进建议。
client-arch-explain
生成客户友好的架构说明（通俗语言解释架构层次、数据流、技术选型），适合演示和评审会议。当准备客户演示或用户验收前使用。
design-consistency-check
检查VS↔SD/DM一致性（value_path是否与序列图/数据模型匹配），生成不一致项清单和修复建议。当Self-Reflection报告一致性问题或准备CONSTRAINT验收前使用。
dev-tech-view-supplement
根据设计补充Dev视角的技术细节（数据库/API/通信协议），帮助Dev快速进入实施。当L1/L2设计缺乏技术细节、或准备向Dev交付前使用。
entry-point-check
验证 VS 的 entry_points 字段是否与 SPEC_PRJ_ENTRYPOINTS（入口类型标签）一致，生成缺失入口清单和补充建议。当准备 CONSTRAINT 验收前使用。
overdesign-detect
检测设计文档中的过度设计（实现细节、技术术语过多），生成过度设计清单和精简建议。当准备CONSTRAINT验收前使用，避免边界合规验收失败。
registry-sync-validator
审计Registry是否与设计artifacts保持一致，生成同步率报告和遗漏清单。当完成设计后、CONSTRAINT验收前使用，避免Registry不一致导致验收失败。
constraints-generator
生成ExecSpecMasterConstraints_L*.md权威约束文件，定义Phase/SubPlan/Round命名规范、目录结构、环境策略和质量门禁阈值。当需要创建 ExecSpec Master Plan 的约束文档、定义命名规范、定义质量门禁、定义HITL检查点时使用。
Page 1191 of 1665 · 83215 results