Inspiration
I'd like to make my mark on history, if that's the case
What it does
NoirGuard is an autonomous DevSecOps agent that audits GitHub pull requests for security vulnerabilities and attempts to remediate them automatically.
It analyzes the full content of changed Python files using an LLM, generates a security patch, and validates the result with pylint and bandit. If the patch fails validation, NoirGuard feeds the structured errors back into the remediation loop and tries again.
It can run as a one-shot CLI audit or automatically react to GitHub pull request events through a webhook. The system also verifies webhook signatures, limits request size and concurrency, and fails closed when security validation cannot produce a reliable result.
The goal is to move beyond simply detecting vulnerabilities: NoirGuard aims to detect, verify, remediate, and re-validate security issues automatically.
How we built it
Well, at first I thought about creating something huge, but then I realized that the simpler, the better
Challenges we ran into
I honestly don't understand how to configure webhooks or what to do with them, but looking at the code, I think everything is correct.
Accomplishments that we're proud of
Well, this is my first time doing a project like this, so I honestly don't know
What we learned
Python writing and some use ai
What's next for NoirGuard
Maybe it'll be a fall, maybe it'll be a rise—I don't know
Log in or sign up for Devpost to join the conversation.