LLM & Agents

Use this whenever a user wants to add a new feature or explitly states to research a feature/API or building a plan for a new feature. It iterviews the user for feature details (if not provided), research the best API/service for their needs, confirm choice, then gather all implementation notes for their request and save them as a .claude/plans file.

Creates structured implementation plans with phase prompts for Claude Code execution. Use when building complex projects, creating implementation roadmaps, breaking work into phases, or generating Claude Code prompts for multi-step development. Triggers include "create implementation plan", "phase this project", "create phases for", "plan the build", "phased implementation", "break this into phases".

Multi-agent code analysis orchestration using claudemem. Share claudemem output across parallel agents. Enables parallel investigation, consensus analysis, and role-based command mapping.

Single-skill vs multi-skill subagent architectures. Use when designing subagents.

This skill should be used when Claude needs to create structured, educational summaries of YouTube videos. Use this skill when the user requests a video summary, content analysis, or learning-focused breakdown of YouTube content. The skill integrates youtube-video-info and youtube-transcript skills to gather video data and transcripts, then generates comprehensive summaries following a specific educational format with key topics, insights, data-based analysis, and exploratory questions.

Automatically runs project validation commands (linters, type checkers, tests) after file edits and blocks commits on any failure. Reads validation commands from CLAUDE.md or auto-detects from project type. Enforces strict quality gates to catch issues early. (project)

Weaviate vector database for semantic search, hybrid queries, and AI-native applications. Use for embeddings storage, similarity search, RAG pipelines, and multi-modal retrieval.

Guide for creating new Claude Code skills following best practices. Use when the user asks to create a new skill, agent capability, or wants to extend Claude's autonomous functionality.

Defines system architecture and technical design decisions

Use this skill when the user wants to call different LLM models through OpenRouter's unified API, compare model responses, track costs and response times, or find the best model for a task. Triggers include requests to test models, benchmark performance, use specific providers (OpenAI, Anthropic, Google, etc.), or optimize for speed/cost.

Discover, browse, and compare agent skills from repositories. Shows new skills, updates, and helps users find relevant skills. Use when exploring available skills or checking for updates.

Guide for creating new skills in @assistantName@'s personal AI infrastructure. Use when user wants to create, update, or structure a new skill that extends capabilities with specialized knowledge, workflows, or tool integrations. Follows both Anthropic skill standards and PAI-specific patterns.

PDF-RAG API reference. REQUIRED after any failed curl/jq to localhost:8000 (404, null, jq error). Also use when uncertain about endpoint path or response shape.

This skill should be used when generating structured metaprompts for repeatable tasks. Triggers include "create a metaprompt for X", "generate a review prompt", "make a prompt template", or when building prompts for cursor-agent, Codex, or other AI tools.

Embed project-local skills, agents, and commands for semantic search. Use after creating items manually or to update embeddings. Works with both keyword and semantic search modes.

Creates new Claude Code slash commands following best practices. Guides through commandstructure, naming, arguments, and frontmatter. Use when user wants to create a command,build a slash command, or asks about command best practices.

Create spike definitions with canonical names and numbered approaches for parallel exploratory implementation. Use when partner has an underdefined feature idea and wants to explore multiple implementation approaches in parallel, when uncertain which technical approach is best, or when comparing alternatives before committing to implementation

Dispatch code-reviewer subagent to review implementation against plan or requirements before proceeding

OpenAI official SDK usage (Python, Node.js). Use when: writing code that calls OpenAI API,implementing chat/embeddings/images/audio features, handling streaming responses,async patterns, error handling with SDK. For raw HTTP/REST calls, see `openai-api` skill.

Master QMIX, MADDPG, CTDE - multi-agent learning with coordination and credit assignment