video1.0<iframe src="https://www.loom.com/embed/5b1dc2dc858b4493b4b348f0b88d5b9e" frameborder="0" width="1910" height="1432" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14321910Loomhttps://www.loom.com14321910https://cdn.loom.com/sessions/thumbnails/5b1dc2dc858b4493b4b348f0b88d5b9e-d11fad52bed75fb1.gif90.923This Loom explains what viewers can see on the local CA web server during an ongoing or successful agent run. It shows the meta agent profile that triggered the run, including the model, agent backend, provider, and the target agent being improved, along with the task it is running. The orchestration loops through each generation by having the meta agent generate agent code, run the full task set, and produce a score, then a feedback agent improves the code for subsequent generations. You can also view the agent code, the prompts, the improvement plan, and the trajectories for individual task questions.