I took some time to test the BMAD agents within the GitHub Mission Control interface. It’s a very promising concept, and I’m a fan of the direction, but it feels like it will need a few more iterations to be truly seamless.
Here are my “at first glance” observations: Performance: The interactions felt quite slow. I noticed it was spinning up a Playwright server, which seemed unexpected for the tasks I was running. Diff View: The ability to see file changes directly in the chat is a great feature. However, it currently lacks more advanced capabilities we’re used to in our editors, like Checkpoints. Model Selection: I didn’t see an option to select the underlying AI model. Having the flexibility to switch models (e.g., based on task complexity, speed, or cost) would be a key addition. UX/Persistence: I couldn’t find a way to pin a specific repository. It would be a significant workflow improvement to have it default to a primary repo when moving between sessions and Mission Control. Am I alone in this? Curious to hear if others have had a similar experience or are finding it valuable in its current state. In the spirit of transparency, I used an AI to help refine the phrasing of this very post—a small but practical example of the value I’ve come to appreciate.
