BeeAI Framework

What it does

• Two parallel libraries, Python and TypeScript, for building production-ready multi-agent systems
• A Workflows module composes multi-agent systems (since January 2025); a Backend module abstracts AI services such as chat and embedding
• MCP tool integration plus ACP support — ACP became part of A2A under the Linux Foundation in August 2025
• An experimental Requirement Agent (Python) constrains agent behavior with declarative requirements

When to use it

Teams that want a Linux Foundation-hosted, Apache 2.0 framework with libraries in both Python and TypeScript for multi-agent orchestration — especially when workflows need to mix custom agents, MCP tools, and A2A/ACP interoperability.

Getting started

Python:

pip install beeai-framework

TypeScript:

npm install beeai-framework

Documentation at framework.beeai.dev/introduction/welcome.

Things to know

The two language libraries are not feature-locked: the README's update log ships Python and TypeScript items on separate dates, so check which language a given module supports before committing. The Requirement Agent is explicitly experimental, and the README's Latest-updates table is maintained irregularly — at fetch time its newest entry trailed the repo's ongoing activity — so use the docs and changelogs for current state.