What is an AI agent?
The loop these tests exercise: a model that calls tools until it can answer.
Learn · Track
How to tell whether an agent works, and whether it still works after you change a prompt, a model or a tool. You should know what an agent is; some experience with automated tests helps.
What an agent eval is: tasks, trials, graders and transcripts, capability vs regression evals, pass@k vs pass^k, and how to write your first 20 cases.
Catch a prompt, model or tool change that breaks your agent before users do: save known-good runs, replay changes against recordings, gate CI.
Why live MCP calls make agent tests flaky, and what recorded replays, schema-built mocks and breaking-change checks each catch and miss.
A release gate lets a workflow version reach production only after its tests pass on that exact version. How gates, drafts, overrides and rollback work.
The loop these tests exercise: a model that calls tools until it can answer.
The step-by-step docs guide: run recorded tests in your pipeline and fail the build on a regression.
This website (not the L2M product) uses Google Analytics, only after you accept cookies. Cookie settings