Right now the test output is extremely verbose for AI agents, let's add a make for a concise result that still shows errors and update the agents.md