Lightweight Python multi-agent framework from CMU and UW-Madison where agents are first-class objects with task delegation, vector stores, tool use, and LLM flexibility.
- Access
- API
- Auth
- API Key
- Skills
- 3
Skills
Agent Task Delegation
Define agents that delegate sub-tasks to specialised child agents and aggregate results via a principled multi-agent task graph.
Agent Tool Use
Equip agents with typed tools using Pydantic schemas so agents can call functions, APIs, or databases and handle responses in a structured loop.
Vector Store RAG
Connect agents to built-in vector store support for document retrieval-augmented generation across LanceDB, Chroma, Qdrant, and others.
Related Agents
OCI Generative AI
API · CLIProvides Oracle Cloud's managed model inference, OpenAI-compatible Responses API, agent tools, vector stores and…
A2A Protocol
APIDefines the Agent2Agent interoperability protocol and SDKs so agents built on different frameworks discover each other…
Flowise
APIOpen-source drag-and-drop UI for building LLM flows and AI agents. Deploy as a self-hosted server with REST API for…
Docker Agent
CLI · MCP · APILets developers define multi-agent teams in YAML and run them with the docker agent CLI, expose them over MCP or an…