(Experiment) Colocating agent instructions with eng docs
In my initial investigation of agent docs (Docs for AI agents) I expressed
skepticism over the current design of agent docs. Agent providers like Claude
Code, Gemini CLI, Cursor, and so on are steering us towards maintaining agent
docs as a new docs set, separate from preexisting internal eng docs. I worry
that we’ll end up duplicating a lot of information across the two docs sets.
Inevitably, the duplicated information gets out-of-sync with each other.
In Colocate, I basically propose embeddin...
Read more at technicalwriting.dev