BRK287 Breakout Fri 20 Nov 09:00 PST
Trusted code at scale, from legacy to AI
Subhash Arja, Eric Tooley
Summary
From the recording
GitHub's Subhash Arja presented GitHub Code Quality, which combines CodeQL rules with AI findings and Copilot agentic autofix, plus pull request merge gates, test coverage and org-level trend reporting. He said it reached general availability in July after a public preview. It needs a separate licence, with a per-committer charge plus a usage-based AI component. Roadmap items include compliance reporting (MISRA, AUTOSAR), test coverage trends and cyclomatic complexity. Caveat: pricing was described as current and under constant evaluation; Azure DevOps availability was not confirmed, and the trial is still being built.
Session description
Microsoft's catalogue description
Every codebase carries quality debt, and AI adds to it faster than teams can pay down. Performance issues inflate cloud cost, maintainability debt slows every future change, and reliability gaps surface as incidents. GitHub Code Quality catches these across new and existing code where developers work, recommends fixes, and gives engineering leaders governance to enforce standards. Agentic autofix then remediates findings at scale, turning backlogs into reviewable pull requests.