aka. Agent Skills
Discover skills for AI coding agents. Works with Claude Code, OpenAI Codex, Gemini CLI, Cursor, and more.
This skill should be used when implementing slash commands that execute without Claude API calls. Use when: adding a new /bumper-* command, understanding why commands return "block" responses, debugging UserPromptSubmit hooks, or learning the pattern for instant command execution. Keywords: UserPromptSubmit, block decision, hook response, slash command implementation.
Surface relevant learnings from past sessions. Use when starting work in a domain to avoid repeating mistakes. Triggers on "recall learnings", "past failures", "surface insights", "what did I learn".
Execute pre-decomposed PRD with machine-verifiable tasks. Triggers on "loop prd", "run prd", "execute tasks from PRD", "prd execution". Parses task blocks with T-NNN format, respects dependencies, supports parallel execution with --parallel flag.
Fresh-instance orchestration for multi-story features. Use when N>5 stories risk context exhaustion. Spawns fresh Claude per story via bash script. Triggers on "ralph", "fresh instances", "multi-story loop", "context exhaustion".
Root cause analysis for complex bugs. Use when initial fix fails, bug keeps returning, or incident is severe. Triggers on "root cause", "why does this keep happening", "trace the bug".
Scaffold structured documents (RFC, ADR, Blog) using industry best practices. Use when creating RFCs, ADRs, blog posts, or technical articles. Triggers on "write RFC", "create ADR", "blog post template", "scaffold document", "RFC template". Produces draft for refinement with /wordsmith:edit.
Bridge WHAT (intent) to HOW (implementation). Use when spec is clear but approach is not. Triggers on "shape this", "how should I build", "implementation approach".
Cancel active Loop. Use when loop needs to stop early, done iterating manually, want to abort, or change direction. Triggers on "cancel loop", "stop iterating", "end loop", "terminate loop", "abort loop", "kill loop", "stop the loop", "halt".
Evaluate skill quality using 8-dimension framework. Use when reviewing skills, before publishing, or auditing existing skills. Triggers on "evaluate skill", "review skill", "skill quality", "audit skills".
Explore visual design, UI/UX, and design systems. Use when designing visual directions, user journeys, information architecture, color palettes, typography, or reviewing design feedback. NOT for feature planning or technical architecture.
Product workflows router. Use when validating product ideas, writing specs, analyzing competitors, or tracking metrics. Triggers on "PRD", "competitive analysis", "metrics", "OKRs", "cohort", "requirements", "tech debt", "game-changing features", market positioning, retention analysis, or product strategy tasks.
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation. Also used during /shipyard:init for requirements gathering.
Use when working with any code, infrastructure-as-code, configuration files, dependencies, or before claiming security posture is adequate — covers OWASP Top 10, secrets detection, dependency vulnerabilities, IaC security, Docker hardening, and supply chain risks
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
Use when creating new skills, editing existing skills, or verifying skills work before deployment
Validate git operations won't affect protected branches or cause data loss
Create file and commit atomically (60-75% faster than step-by-step)
Guide for writing clear, descriptive commit messages
Rewrite git history using git-filter-repo (not filter-branch)
Find and read multiple files in one operation (50-70% faster for exploration)