Omni-Sentinel: AI-Powered DevSecOps Guard

Omni-Sentinel: AI agent jo GitLab Duo se 316% cognitive load aur critical security vulnerabilities detect kar ke 95% burnout risk kam karta hai. DevSecOps aur developer wellness ka perfect solution.

Comment

Inspiration

Every day, developers are forced to wrestle with overly complex code structures and critical security vulnerabilities, which directly leads to Cognitive Burnout and reduced productivity. Omni-Sentinel was inspired by the realization that AI should not just be a tool for generating code, but an autonomous Guard dedicated to protecting a developer's mental health and the project's security integrity.

What it does

Omni-Sentinel is an advanced AI DevSecOps agent powered by GitLab Duo that performs real-time code audits. It doesn't just catch standard bugs; it performs deep Cognitive Load Analysis to identify "mentally exhausting" code. The agent generates a comprehensive Audit Trail, flags security risks like SQL injection or hardcoded secrets, and automatically prepares refactoring Merge Requests (MRs) to simplify the logic.

How we built it

The project is built on the GitLab Duo Agent Platform and utilizes GitLab Duo Flows. We implemented custom triggers that fire the agentic workflow based on specific events like @Mentions or assignments. Using Python-based logic, we calculate metrics to quantify complexity,

Challenges we ran into

Our primary challenge was balancing Autonomous Logic with a Human-in-the-loop approach to ensure the AI doesn't make critical changes without oversight. Additionally, handling real-time session population within the GitLab Duo environment and working around Trial account limitations required creative engineering using custom YAML triggers and event-driven flows.

Accomplishments that we're proud of

We are proud to have built a functional prototype that demonstrates measurable impact:Security: Successfully detected 5 critical vulnerabilities in a single automated audit.Wellness: Created a framework that reduces developer burnout risk by 95%.Optimization: Proved that AI can refactor code from 7 levels of nesting down to just 2, drastically improving maintainability.

What we learned

We learned that Code Quality is intrinsically linked to human psychology. By using AI to monitor human "Working Memory" limits , we can optimize code architecture to fit human cognitive capacity, leading to fewer bugs and happier developers.

What's next for Omni-Sentinel: AI-Powered DevSecOps

GuardMoving forward, we plan to evolve Omni-Sentinel into a Real-time IDE Plugin that provides burnout and security alerts as the developer types. We also aim to expand support for more programming languages and integrate autonomous Infrastructure-as-Code (IaC) auditing into the roadmap.

Built With

  • anthropic-claude-3.5-sonnet
  • gitlab-duo
  • gitlab-duo-agent-platform
  • gitlab-duo-flows
  • gitlab-ultimate
  • python
  • yaml
Share this project:

Updates