Ask HN: Is it better to have no Agent.md than a bad one?

https://news.ycombinator.com/item?id=47124103

by parvardegr • 5 months ago

10 7 5 months ago

news.ycombinator.com

Please share your real word experiences. What is a bad one and why?

verdverm2• 5 months ago

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.

parvardegr2• 5 months ago

Hmm sounds reasonable

verdverm2• 5 months ago

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

nullcathedral• 5 months ago

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...

parvardegr2• 5 months ago

Thanks for this

e5bed97b3349• 5 months ago

Yes, zero instructions are better than wrong instructions because outdated or incorrect context actively forces the AI to write broken code.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

DANmode• 5 months ago

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

Ask HN: Is it better to have no Agent.md than a bad one?

Related Stories