# BeeAI Framework

> An open-source framework for building production-ready multi-agent systems in Python and TypeScript, hosted by the Linux Foundation under Apache 2.0.

BeeAI Framework is an open-source toolkit for building production-grade multi-agent systems, available in both Python and TypeScript with complete feature parity. Developed by contributors to the BeeAI project and hosted by the Linux Foundation AI & Data program, it provides a lightweight yet powerful approach to reliable agent development. The Python alpha launched in February 2025, and the project has accumulated over 3,200 GitHub stars as of mid-2026.

## What It Is

BeeAI Framework is a developer library that gives teams everything needed to create intelligent, autonomous agents and orchestrate them into multi-agent systems. It goes beyond simple LLM prompting by providing built-in constraint enforcement and rule-based governance that preserves reasoning abilities while ensuring predictable, deterministic behavior. The framework is provider-agnostic, supporting 10+ LLM providers including Ollama, Groq, OpenAI, and Watsonx.ai.

## Core Architecture and Modules

The framework is organized into composable modules that cover the full agent development lifecycle:

- **Agents** — Create intelligent agents with the `RequirementAgent` class, which enforces rules the agent must follow for controlled, predictable behavior across different LLMs.
- **Backend** — Unified interfaces for connecting to any LLM provider with seamless switching.
- **Tools** — Built-in tools (web search, weather, Wikipedia, code execution) plus support for custom tools and MCP-compatible tools.
- **RAG** — Retrieval-augmented generation with vector stores and document processing.
- **Memory** — Built-in memory strategies for managing conversation history.
- **Workflows** — Orchestrate multi-agent systems with advanced patterns like parallelism, retries, and replanning using simple decorators or declarative YAML.
- **Serve** — Host agents in servers with support for A2A and MCP protocols.
- **Cache** — Intelligent caching to optimize performance and reduce costs.
- **Serialization** — Save and load agent state for persistence across sessions.
- **Observability** — Native OpenTelemetry support for real-time monitoring, auditing, and tracing.

## Protocol Integrations

BeeAI Framework is natively compatible with both the Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocol. The MCP integration, added in May 2025, allows agents to be equipped with MCP tools and to expose themselves as MCP-compatible components. The A2A protocol integration enables interoperability with any A2A agent system. Notably, the project's README states that ACP (Agent Communication Protocol) became part of A2A under the Linux Foundation in August 2025.

## Setup Path

Installation is straightforward via standard package managers:

- Python: `pip install beeai-framework`
- TypeScript: `npm install beeai-framework`

Starter templates (`beeai-framework-py-starter` and `beeai-framework-ts-starter`) are available on GitHub to accelerate onboarding. The framework supports local inference via Ollama, making it possible to run fully local multi-agent pipelines without cloud dependencies.

## Update: typescript_v0.1.29

The latest release is `typescript_v0.1.29`, published May 5, 2026. Recent notable additions include the experimental Requirement Agent (June 2025), ACP/MCP protocol integrations (May 2025), a Backend module for simplified AI service access (February 2025), Workflows for multi-agent orchestration (January 2025), and support for DeepSeek R1 and LLaMA 3.3. The project is actively maintained with 438 forks and 18 open issues as of mid-2026, and follows open, community-driven governance under the Linux Foundation.

## Features
- Multi-agent system orchestration
- Python and TypeScript support with feature parity
- RequirementAgent with constraint enforcement
- Built-in tools (web search, weather, Wikipedia, code execution)
- RAG with vector stores and document processing
- Conversation memory management
- Dynamic workflows with parallelism, retries, and replanning
- Declarative YAML orchestration
- MCP and A2A protocol native support
- 10+ LLM provider support (Ollama, Groq, OpenAI, Watsonx.ai, etc.)
- Native OpenTelemetry observability
- Intelligent caching
- Agent state serialization and persistence
- Agent serving with multi-protocol support
- Experimental Requirement Agent for predictable behavior

## Integrations
Ollama, OpenAI, Groq, Watsonx.ai, DeepSeek R1, LLaMA 3.3, Wikipedia, OpenMeteo, Model Context Protocol (MCP), Agent-to-Agent (A2A), OpenTelemetry, Streamlit

## Platforms
LINUX, API, DEVELOPER_SDK, CLI

## Pricing
Open Source

## Version
typescript_v0.1.29

## Links
- Website: http://framework.beeai.dev
- Documentation: https://framework.beeai.dev/introduction/welcome
- Repository: https://github.com/i-am-bee/beeai-framework
- EveryDev.ai: https://www.everydev.ai/tools/beeai-framework