One approach (Claude Code) is to evolve it over time. Start small and run /insights often and use that to refine the CLAUDE.md as needed.

https://github.com/trailofbits/claude-code-config?tab=readme...

Thanks for this

My approach is to keep them minimal. Make them a table of contents and spark notes, spread them across the repo, keep them up to date as the repo changes.

Something is better than nothing. It's pretty hard to have bad ones, also takes effort to craft good ones. The bad ones are the ones with incorrect information.

Hmm sounds reasonable

If paired this with system prompts derived from Claude Code. This makes gemini-flash nearly on par with the big boy models. Saves a ton of time and money

Yes, zero instructions are better than wrong instructions because outdated or incorrect context actively forces the AI to write broken code.

Start with nothing, and only add a single sentence to your Agent.md after the AI repeatedly makes the same specific mistake.

Without further instruction, it falls back to the default logic.