Harbor (Agent Evaluation Framework)

Harbor is an open-source framework, currently at version 0.3.0, designed for evaluating and optimizing AI agents and language models using sandboxed environments. It facilitates the creation and execution of benchmarks, allowing users to assess arbitrary agents and models. The library is actively developed, with its current release focusing on providing tools for robust and scalable agent evaluation. Release cadence is not explicitly stated but updates appear to be regular.