Resources

CrewAI APIs and the CLI move quickly — prefer the official docs. This page lists primary docs, related courses on this site, and coding-agent skills.


Official

Coding agents:

npx skills add crewaiinc/skills

CourseWhen to open it
LangChainSingle-agent harness, middleware, LangGraph graphs and checkpoints
LlamaIndexDocument indexes / query engines; wrap with LlamaIndexTool
RAGChunking, embeddings, Chroma / pgvector / Qdrant, evaluation
MCPAttach MCP servers as agent tools
OllamaLocal models via CrewAI provider/model-id
DifyVisual workflows / agents vs code-first CrewAI

Tutorial index

  1. CrewAI Tutorial
  2. Introduction
  3. Installation
  4. Quick Start
  5. Agents
  6. Tasks
  7. Crews and Processes
  8. Tools
  9. Flows
  10. Memory
  11. Practical Examples

CLI cheat sheet

CommandPurpose
uv tool install crewaiInstall the global CLI
crewai create crew <name>JSON-first Crew project
crewai create crew <name> --classicYAML + crew.py
crewai create flow <name>Flow project (official Quickstart)
crewai install / crewai runInstall deps and run the entrypoint
npx skills add crewaiinc/skillsOfficial skills for coding agents

Model IDs are provider/model-id (e.g. openai/gpt-4o-mini). Tools extra: pip install 'crewai[tools]' or uv add 'crewai[tools]'.

Python version follows the install page: currently >=3.10 and <3.14. Flow decorators follow the concepts page: @start, @listen, @router, or_, and_, @persist, @human_feedback. Classic Crew scaffolds also use @CrewBase / @agent / @task / @crew — do not put those on a Flow class.


Contribute

Part of Kenhuang Academy. When upstream APIs change, open an issue or PR and update both docs/zh/crewai/ and docs/en/crewai/.

评论