crewai-multi-agent-frameworklisted

# CrewAI Multi-Agent Framework ## When to activate Building role-based multi-agent workflows in Python; user mentions CrewAI; need a working multi-agent prototype fast; the problem maps naturally to job roles (researcher, writer, reviewer, analyst); sequential pipeline logic with Claude as the underlying model. ## When NOT to use TypeScript projects — use Mastra instead; complex conditional routing or branching logic required — use LangGraph; production systems that need checkpointing, failure recovery, and resumable runs — use LangGraph; single-agent tasks where the role abstraction adds overhead without benefit. ## Instructions **Core model:** CrewAI organizes work as a `Crew` of `Agents` running `Tasks` with `Tools`. The framework handles agent-to-agent communication, output passing, and process orchestration. **Installation:** ```bash pip install crewai crewai-tools ``` **Three core concepts:** - `Agent` — a role with a goal, backstory, LLM, and tool list. Defines who does the work. - `Task` — a description, expected output, and assigned agent. Defines what gets done. - `Crew` — the collection of agents and tasks with a process type. Defines how it runs. **Process types:** - `Process.sequential` — tasks run in order, each output available to the next (default, simplest) - `Process.hierarchical` — a manager agent reads all tasks and delegates to specialist agents dynamically **Claude integration:** ```python from langchain_anthropic import ChatAnthropic llm = Chat