Consensus on using actual cursor rules vs `./docs/` folder

Discussions
,
1

Hi, in my current workflow I like using Cursor’s agents and Claude code in a terminal as well and i think this is perfect for my use case and budget.

However I’m trying to optimise my agents and rules without having to duplicate or symlink everything, is .cursor/rules really the best way to do get it to follow them?

Can i just do progressive disclosure by simply having a docs/ folder and pointing to it from my thin AGENTS.md?

1 Like
3

I use Cursor and OpenCode, so this might be related.

For global rules, I set them up manually in both Cursor and OpenCode. This includes things like the agent’s response harness and how it plans its steps.

For project scope, I configure it in Cursor. I then let agents use it when relevant, or I mention it explicitly in Cursor or OpenCode.

For docs, I usually use them for documentation such as PRDs, RFCs, and other technical or brainstorming materials.