We believe sometimes its worth spending more time aligning on agent goals & evals before coding for e.g. spec alignment, generating evals, provisioning sandboxes, etc.

Right now our package only works for langchain agents & supports a few MCPs. Roasts are welcome.