Any-agent: A single interface to use and evaluate different agent frameworks

3 months ago 3

Why any-agent?

any-agent simplifies the process of building agents and testing agent frameworks. No matter which framework you use, the trace formatting looks the same and follows GenAI open telemetry standards. It lets you compare different agent setups in a consistent way, so you can easily figure out why your agents fail when they do.

Supported frameworks: Agno, Google ADK, LangChain, LlamaIndex, OpenAI Agents SDK, Smolagents and TinyAgent

One Interface many Frameworks

Use a single interface for multiple frameworks. Build once and switch with a simple configuration change, no need to rewrite code.

Built-in Tracing and Evaluation

any-agent standardizes tracing across frameworks and provides 'trace-first' evaluation methods, making comparing agent frameworks easier than ever.

Support for MCP and Agent2Agent

Enables integration with MCP and serving of A2A, making it easy to build powerful and connected agent systems

Get Started with any-agent

Pick a framework, set up a config, and try out one of the examples.The documentation has everything you need to go further.

See it in action

Explore this surf spot finder our team built using any-agent:

Sign up for our updates

We regularly share articles on our projects, learnings and collaborations. Sign up to be the first to know.

Read Entire Article