Chrome for Developers: How to evaluate your WebMCP tools

Are your AI agents using your WebMCP tools correctly? Learn how to write evaluations or "evals" to test non-deterministic agent behaviors that traditional unit tests can't catch. Discover how to use the open-source webmcp-evals CLI to evaluate tool selection, review agent trajectories, and debug user journeys across repeated runs. Check out the links below to get started, and share your feedback in the comments!



Resources:

WebMCP Evals on GitHub → https://goo.gle/webmcp-evals-cli

WebMCP Evals on npm→ https://goo.gle/webmcp-evals-npm

WebMCP documentation → https://goo.gle/webmcp-docs

WebMCP and agents → https://goo.gle/4k4TxDx



Subscribe to Chrome for Developers → https://goo.gle/ChromeDevs



#Chrome #ChromeforDevelopers #WebMCP



Speakers: Maud Nalpas

Products Mentioned: WebMCP, Web agents