Ecoaudit is an open-source tool that finds avoidable compute waste using rule-based checks, with no LLM or machine learning. It reads code from a public GitHub repo or a local Git checkout without running it. You can add existing deployment logs through read-only AWS IAM access and artifacts such as profiles or test reports from runs in your own environment. Ecoaudit checks that evidence against its rules, and you can run the entire app on localhost to keep your code and artifacts private.
Open live project- Performance profiling: Built a GitHub Action that collects Python CPU and memory profiles and runs Lighthouse, then uploads the results for analysis.
- Database checks: Added detectors that flag wasteful SQL patterns and query plans, including sequential scans inside nested loops.
- Report storage: Updated the Lighthouse worker to save JSON and HTML reports to private S3 and added a live demo script.