Agent to assess security issues in context, and auto-fix

Feature Requests
1

Can you please provide a background agent that assesses security issues in context, and auto-fixes them?

Problem:
When I run npm audit, it shows me security risks, but I have these problems with it:

  • poorly formatted
  • to get extra context I need to go to a website on the given issue
  • while an issue may be marked ‘critical’, it may not be to me (for example if the criticality doesn’t apply if you host on AWS)

Solution:
Cursor background agent.
Every week it runs my .mdc rule which says something like:

  • Run npm audit
  • Investigate each issue in the context of codebase and implementation
  • Where you believe issues should be fixed, create a PR for each one, starting with highest priority.
2

Is that not already available using Background Agents?
You can assign it the task as you described.

One thing will be hard for AI to know, what is critical for you and what isnt unless you specify the rules.

3

Ah, that’s great if it will support it then.

I haven’t used it because it hasn’t been security audited itself yet.

But I will when it has.