PerformanceSentinel

Every merge request today is reviewed for correctness but almost never for performance impact at scale. That’s the gap we’re solving.

I built Performance Sentinel, an AI-powered GitLab workflow agent that automatically analyzes every merge request for hidden performance regressions before they hit production. It doesn’t just scan code it understands diffs, inspects changed files, and predicts how small inefficiencies can amplify under real world load. For example, a seemingly harmless 100ms delay can translate into measurable CO₂ emissions at scale, so the agent quantifies environmental impact alongside performance.

What makes this unique is that it’s fully integrated into the developer workflow no extra tools, no friction. The agent comments directly on the merge request with actionable insights, helping engineers optimize early rather than firefight later.So instead of reacting to performance issues in production, teams can proactively ship code that is faster, more scalable, and more sustainable by design

Challenges we ran into

One of the biggest challenges was reliably extracting and understanding code changes from each merge request. Identifying meaningful performance issues without producing false positives was difficult, especially when the context of the codebase is limited. The agent sometimes struggled to distinguish between intentional design decisions and actual inefficiencies. Ensuring consistent and accurate analysis across different coding styles and structures was another hurdle. Additionally, scaling this approach beyond small projects proved challenging, as larger codebases introduce complexity in dependency tracking, context understanding, and performance estimation.

What's next for Performance Sentinel

We plan to expand Performance Sentinel to handle large-scale and legacy codebases, where performance issues are often deeply embedded. The goal is to help teams not just catch new regressions, but also gradually clean and optimize existing code. We aim to improve context awareness by incorporating historical data and deeper codebase understanding. Future iterations will also focus on more accurate performance modeling, better handling of complex architectures, and broader support across different project sizes and tech stacks.

Built With

• claude
• gitlab
• python