# Grok Bot Token usage **URL:** \ **Category:** Help\ **Tags:** grok-bot\ **Created:** [October 14, 2026, 11:44am UTC](https://forum.cursor.com/t/grok-bot-token-usage/174723 "2026-10-14T11:44:24Z")\ **Posts on this page:** 1\ **Showing post:** 17
**Author:** ![Colin](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/colin/32/90452_2.png) [@Colin](https://forum.cursor.com/u/Colin)\ **Post date:** [October 15, 2026, 10:08am UTC](https://forum.cursor.com/t/grok-bot-token-usage/174723/17 "2026-10-15T10:08:28Z")
Hey @AndrewC, yes, that exists. Cloud agents read a file at `.cursor/CLOUD.md` (inside the `.cursor` folder at the root of the repo, not a `CLOUD.md` at the root itself). It is only loaded for cloud agents, so agents in the IDE never see it. That makes it a good home for cloud-only notes: how to start services, how to run tests, and any gotchas on the cloud machine. The approach in our docs is a section in your root AGENTS.md titled `## Cursor Cloud specific instructions`, which cloud agents also pick up: [Cloud Environment Setup | Cursor Docs](https://cursor.com/docs/cloud-agent/setup) The difference is that AGENTS.md is read by every agent, IDE included, so anything in that section also lands in your local agents’ context. A good split: keep the app map (main folders, where key logic lives, conventions) in AGENTS.md, since every agent benefits from it, and put cloud-only setup and testing notes in `.cursor/CLOUD.md`. Keep both short and link out to other files for detail. Whatever is in them gets re-read at the start of every run, so a lean file is also cheaper. --- _[View the full topic](https://forum.cursor.com/t/grok-bot-token-usage/174723)._