I have defined a set of instructions in my rules.md file inside .cursor/rules folder. One of the items i had mentioned is to never mock data. There are few other rules, but cursor seems to add mock data when the real server/integration is not working.
Isnt every chat session in the cursor IDE follow the rules defined in .cursor/rules folder
Steps to Reproduce
Create a rule.md file inside .cursor/rules folder. Set an instruction stating that no mock data be used/generated
Expected Behavior
No mock data being created
Operating System
MacOS
Current Cursor Version (Menu → About Cursor → Copy)
Hey, thanks for the report. Lately, we’ve seen several reports that rules aren’t being applied correctly.
To help narrow it down, please share:
Which model you’re using in chat (Claude Sonnet, GPT, Cheetah, etc.)?
Where this happens: Chat, Agent, custom mode, or Background Agent?
What to do:
Use the .mdc extension: in .cursor/rules/rules.mdc add conditions with alwaysApply: true or autoAttach: true. Rules | Cursor Docs
Check whether the rules show as active in the Agent sidebar when you’re in a chat session
Temporary workarounds:
Reference your rules with @Rules in each chat
Add a “no mock data” instruction directly to your request
Set “Always Apply” in the rule if it’s not already
There are known issues with rules not loading in Background Agent and some other contexts, so this info will help determine if it’s the same issue or something different.