Total 31,934 skills, AI & Machine Learning has 5154 skills
Showing 12 of 5154 skills
Collaborative coding with enforced micro-steps: announce, show diff, wait for confirmation, apply, verify. User controls pace with commands. Works with any domain agent as the executor. Use when: "pair program", "pair with me", "let's code together", "step by step coding", "walk me through implementing", "code with me"
RED-GREEN-REFACTOR testing for agents: dispatch subagents with known inputs, capture verbatim outputs, verify against expectations. Use when creating, modifying, or validating agents and skills. Use for "test agent", "validate agent", "verify agent works", or pre-deployment checks. Do NOT use for feature requests, simple prompt edits without behavioral impact, or agents with no structured output to verify.
Restore session state from handoff artifacts and route to the next action. Priority cascade: HANDOFF.json (highest) > .continue-here.md > incomplete task_plan.md > git log. Presents a status dashboard, then executes the next action. Use for "resume", "continue", "pick up where I left off", "what was I doing", "continue work". Do NOT use for starting new tasks (use /do), reviewing past sessions (use /retro), or reading task plans (read task_plan.md directly).
CLI-based image generation from text prompts using Google Gemini APIs via Python. Use when user needs "generate image", "create image with AI", "gemini image", "text to image", "create sprite", or "generate character art". Supports model selection, batch generation, watermark removal, and background transparency. Do NOT use for web app image features (use nano-banana-builder), video/audio generation, or non-Gemini models.
Weighted decision scoring framework for architectural and technology choices. Frames decisions with 2-4 options, scores against weighted criteria, detects close calls, and records decisions in the active ADR or task plan. Use when: "should I use X or Y", "which approach", "compare options", "trade-offs between", "help me decide", "evaluate alternatives"
Maintain /do routing tables and command references when skills or agents are added, modified, or removed. Use when skill/agent metadata changes, after skill-creator-engineer or agent-creator-engineer runs, or when routing tables need synchronization. Use for "update routes", "sync routing", "routing table", or "refresh /do". Do NOT use for creating new skills/agents, modifying skill logic, or manual /do table edits.
Spawn 10 independent parallel agents to analyze source material from distinct perspectives, synthesize findings, and apply improvements to a target agent or skill. Use when source material is complex and multi-angle extraction justifies 3-5x token cost over inline analysis. Use for "parallel analysis", "multi-perspective", or "deep extraction". Do NOT use for routine improvements, simple source material, or when token budget is limited.
Constructive critique through 5 HackerNews commenter personas with evidence-based claim validation. Use when user wants devil's advocacy, stress testing, or critical review of ideas, docs, architecture, or code. Use for "roast", "critique this", "poke holes", "devil's advocate", "stress test", or "what's wrong with". Do NOT use for code review (use systematic-code-review), implementation changes, or performance profiling without a specific critique request.
Multi-step voice content generation with deterministic validation. Orchestrates a 7-phase pipeline: LOAD, GROUND, GENERATE, VALIDATE, REFINE, OUTPUT, CLEANUP. Use when generating content in a specific voice, writing as a persona, or validating existing content against a voice profile. Use for "voice write", "write as", "generate in voice", or "voice content". Do NOT use for creating new voice profiles (use voice-calibrator), analyzing writing samples (use voice_analyzer.py), or general content without a voice target.
Execute wave-ordered implementation plan by dispatching tasks to domain agents. Use after /feature-plan produces a plan. Use for "implement feature", "execute plan", "start building", or "/feature-implement". Do NOT use without a plan or for ad-hoc coding tasks.
Fresh-subagent-per-task execution with two-stage review (ADR compliance + code quality). Use when an implementation plan exists with mostly independent tasks and you want quality gates between each. Use for "execute plan", "subagent", "dispatch tasks", or multi-task implementation runs. Do NOT use for single simple tasks, tightly coupled work needing shared context, or when the user wants manual review after each task.
Question-only debugging mode that guides users to find root causes themselves through structured questioning. Never gives answers directly. Escalates to systematic-debugging after 12 questions if no progress. Use when: "rubber duck", "help me think through this bug", "debug with me", "walk me through debugging", "socratic debug", "think through this issue"