agentbench — The Regression Testing Framework for AI Agents. Replay · Evaluate · Assert · Catch Regressions — in CI. Like Jest for yo

AgentBench is a new open-source regression testing framework for AI agents. It enables developers to replay agent interactions, evaluate performance, and assert outcomes, integrating directly into CI/CD pipelines. This tool aims to provide a Jest-like experience for the AI layer, helping catch regressions before they impact production.

Opening Kapyn…