What problem does it solve? When a project brief explicitly requires CrewAI for multi-agent orchestration, this Skill ensures the implementation uses CrewAI's actual current API rather than outdated or hallucinated patterns, since CrewAI's constructors and process types change across versions. ## Core Features & Use Cases - Research-first verification: Fetches CrewAI's current official quickstart docs and diffs them against the bundled reference before writing code. - Known-working reference crew: Provides references/basic_crew.py, a minimal 2-agent sequential crew (researcher + writer) with a clean run_crew(input) entrypoint. - Provider-agnostic LLM wiring: Routes all LLM calls through the project's llm_client.py so whichever real provider was verified by the API-key gate is used, with no mock mode. - Use Case: A course brief requires hands-on CrewAI for a multi-agent demo. The Skill scaffolds the crew under concepts/<slug>/agent/, spike-tests it standalone against the installed CrewAI version, then integrates it into the project. ## Quick Start Ask the agent to build the CrewAI multi-agent component for the project whose design document explicitly names CrewAI.