Cursor ignores project rules

Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Per the documentation: “Large language models don’t retain memory between completions. Rules provide persistent, reusable context at the prompt level.”

This sounds like a great way to get around one of the pitfalls of Cursor and LLMs, but when I created a project rule, Cursor ignored it. I know it ignored it, because I asked the agent and it said “I haven’t been following the rules. Apologies.”

Then it said it would read them in the future before each prompt and follow the rules, and almost immediately violated it. It appears that Cursor is relying on the model remembering that it needs to read the rule, and since “large language models don’t retain memory between completions,” it doesn’t remember that it needs to read and follow the rules.

This really needs to be fixed!

Steps to Reproduce

  1. create a .mdc rule in .cursor/rules
  2. start using Cursor agent
  3. note that while it seems to follow rules for new chats, it forgets about them after a few messages, making rules more or less useless

Expected Behavior

Cursor rules are not any better than asking an agent to “always do ___” in chat - they might apply sometimes, but they frequently “forget” that they are always supposed to be run.

Operating System

Windows 10/11

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.0.43 (user setup)
VSCode Version: 1.99.3
Commit: 8e4da76ad196925accaa169efcae28c45454cce0
Date: 2025-10-30T18:49:27.589Z
Electron: 34.5.8
Chromium: 132.0.6834.210
Node.js: 20.19.1
V8: 13.2.152.41-electron.0
OS: Windows_NT x64 10.0.26100

For AI issues: which model did you use?

Composer 1, Claude-sonnet-4.5

Additional Information

It has been an issue for at least 7 months, as evidenced by this thread: Cursor just ignores rules

Does this stop you from using Cursor

No - Cursor works, but with this issue

2

Agreed. Cursor/LLM is doing whatever it wants. If it will not adhere you the rules, then how can you trust the output. If you cant trust the output what good is it. I am discouraged as I have not seen any Cursor representitives giving this the attention it needs.

3

I agree. I don’t know why the cursor doesn’t fix the rules.
They don’t fix it, they can’t fix it, or they’re going to fix it. No answers whatsoever.

4

hi @Kinland @jee1lee @user1094 and thank you for the detailed bug report.

AI models are not deterministic, therefore when they processes the rules but may ignore them. This is not necessarily a Cursor thing but rather LLM related.

Note that AI also tries to be helpful so when you ask it if it followed the rule it may just reply that it didn’t and apologize as well as promise it will never forget to follow the rules again.

There are best practices that you can do to ensure AI follows the rules

  • Keep Agent threads short and focused on single tasks.
  • Ensure that rules are not too long or overly complex.

Feel free to share one example rule it did not follow I will check if there are best practices that I can recommend.

5

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.