If you're building multi-agent systems where different AI personas need to collaborate, this is the framework to reach for. You define agents with specific roles and backstories (researcher, analyst, writer), create tasks with clear outputs, and orchestrate them in sequential or hierarchical workflows. The YAML config approach keeps things clean, and the hierarchical mode gives you a manager agent that delegates work. It shines for structured pipelines like research-to-content workflows but can feel heavy if you just need a simple chain. The planning feature is neat for complex jobs where agents benefit from seeing the overall execution plan upfront. Works best with 3-5 agents, not dozens.
npx -y skills add davila7/claude-code-templates --skill crewai --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills