Beeai Framework

BeeAI Framework is an open source framework for building production grade systems of multiple cooperating AI agents in Python or TypeScript. It is hosted by the Linux Foundation under open governance and focuses on reliability: rather than relying on prompting alone, it lets developers enforce deterministic rules on agent behavior while preserving the model's ability to reason. It is for engineering teams that need agents to behave predictably in production. Agents with constraints apply rule based governance instead of merely suggesting behavior. Dynamic workflows use simple decorators to express patterns such as parallelism, retries and replanning, and declarative orchestration lets teams define whole agent systems in YAML. Production features include built in caching, memory optimization and resource management, and native OpenTelemetry support provides monitoring, auditing and detailed tracing inside an existing observability stack. The framework is native to both MCP and A2A, so agents can use MCP tools, expose MCP compatible components and interoperate with other MCP or A2A agent systems. It supports more than ten LLM providers, including Ollama, Groq, OpenAI and watsonx.ai, and developers can switch between them without rewriting their agents. The framework is free and open source, and no paid tier is published. Costs come from the model providers and infrastructure a team chooses to run it on. The Python and TypeScript implementations have complete feature parity, so mixed language teams can build on the same concepts, and community support runs through Discord, with tutorials on YouTube.