Bug manager: interactive demo.
Run a simulated scan of twelve fictional repositories, watch the agent open pull requests for the fixes, then explore the ranked findings and burn the list down in a sprint.
Simulated data. Every repository, finding, and ticket here is generated in your browser. No code is scanned, nothing is sent anywhere, and no language model is called. The real system ran Claude agents over real repositories, and its findings are confidential, so this demo replays the workflow with made-up ones.
- ScanAgents read every repository
- ClassifyEach finding gets a category
- RankHighest risk first
- Log to JiraOne ticket per finding
- Fix and open a PRThe agent proposes a fix
- Review and burn downEngineers review the PRs
Scan the repositories.
A ranked list, not a pile.
Run the scan to generate the findings.
Spend one sprint reviewing from the top of the list.
Run the scan first, then come back to burn the list down.
━ Fixed in ranked order ╍ Fixed in no particular order
What is real, and what is illustrative.
The workflow is real: scan every repository, classify what is found, rank it, log a ticket for each finding, have the agent attempt a fix and open a pull request, and give a team a time-boxed sprint to review and burn the list down. In the real list, each ticket linked straight to its pull request so an engineer could review it in a click. Everything on this page is simulated, including the code, the findings, the pull requests, and the ticket text.
In the demo, the agent does not open a pull request for every finding: when its confidence is low, or the fix needs a design decision, the finding is flagged for an engineer instead. That rule is illustrative.
The ranking here is illustrative. Each finding’s risk score is its severity weight, multiplied by how exposed clients are to that repository, multiplied by the agent’s confidence. Open a finding to see its score worked out.
The mix of findings mirrors what the real scans surfaced most: data-consistency problems and code paths that can fail on a missing value, with fewer security findings and code smells.