Yesterday I heard about lat.md [1] which seems to have similar ideas about annotating code with spec refs. I now need to try them both.

[1] https://www.lat.md/

Oh cool, thanks for sharing. LLMs tend to adopt these patterns on their own, so I'm sure my ideas are far from unique.

Agents.md actually scales and outperforms alternatives so I would be skeptical about overcomplicating the proces https://vercel.com/blog/agents-md-outperforms-skills-in-our-...