> Your codebase becomes a reinforcement learning environment.
Because the AGENTS.md files (sorry "intent modules" or whatever term you used) change their content based on git diffs? Like really? How is the codebase "learning" anything? And by that logic, aren't all well-maintained repos already "reinforcement learning environments"?
The words and anthropomorphizing is light in this one compared to others, but don't get hung up on that
Adding agent.md files even to good, high quality code bases is helpful. Make it minimal, spark note / table of contents, very basics... Put them in many places in your repo, and then watch to see how much better they are.
Auto injecting high level overviews helps a ton, because every time is basically from a blank slate without them
> Your codebase becomes a reinforcement learning environment.
Because the AGENTS.md files (sorry "intent modules" or whatever term you used) change their content based on git diffs? Like really? How is the codebase "learning" anything? And by that logic, aren't all well-maintained repos already "reinforcement learning environments"?
The words and anthropomorphizing is light in this one compared to others, but don't get hung up on that
Adding agent.md files even to good, high quality code bases is helpful. Make it minimal, spark note / table of contents, very basics... Put them in many places in your repo, and then watch to see how much better they are.
Auto injecting high level overviews helps a ton, because every time is basically from a blank slate without them