any-agent
A single interface to use and evaluate different agent frameworks
...The library is designed to simplify experimentation and benchmarking, making it easier to compare agent behavior, performance, and reliability across different implementations. It includes built-in evaluation tools such as LLM-as-a-judge and agent-as-a-judge methodologies, which help developers assess output quality systematically. any-agent also supports advanced capabilities like tracing, callbacks, and standardized observability through OpenTelemetry, ensuring consistent monitoring across frameworks. ...