{"slug":"cloudflare-runs-a-multi-agent-ai-code-reviewer-on-merge-requests-across-5-169-repositories","url":"https://findausecase.com/use-cases/cloudflare-runs-a-multi-agent-ai-code-reviewer-on-merge-requests-across-5-169-repositories","title":"Cloudflare runs a multi-agent AI code reviewer on merge requests across 5,169 repositories","description":"Cloudflare built a CI-native AI code review system around the open-source OpenCode agent, in which a coordinator agent manages up to seven specialised reviewers covering security, performance, code quality, documentation, release management and internal Engineering Codex compliance, with risk tiers, circuit breakers and model failback chains. Per the source, between 10 March and 9 April 2026 it completed 131,246 review runs across 48,095 merge requests, with a median review time of 3 minutes 39 seconds and an average cost of $1.19 per review.","company":"Cloudflare","industry":"Technology & Software","aiCapabilities":["Agentic AI","Generative AI"],"deployment":"Unknown","problemStatement":"Code review is a good mechanism for catching bugs and sharing knowledge but can bottleneck an engineering team; across Cloudflare's internal projects the median wait for a first review was often measured in hours. Off-the-shelf AI review tools did not offer enough flexibility for an organisation of Cloudflare's size, and a naive prompt over a git diff was noisy.","solutionApproach":"Cloudflare built a CI-native orchestration system around OpenCode, an open-source coding agent. A coordinator agent manages up to seven specialised reviewers covering security, performance, code quality, documentation, release management and compliance with the internal Engineering Codex, deduplicates their findings, judges severity and posts a single structured review comment. Merge requests are classified into risk tiers, and circuit breakers and model failback chains handle provider failures.","businessValue":"Cloudflare reports that in the first 30 days (March 10 to April 9, 2026) the system completed 131,246 review runs across 48,095 merge requests in 5,169 repositories, with a median review time of 3 minutes and 39 seconds and an average cost of $1.19 per review. Engineers used the 'break glass' override on 288 merge requests (0.6%). These are company-reported figures.","evidence":{"band":"high"},"sourceUrl":"https://blog.cloudflare.com/ai-code-review/","dates":{"publishedAt":"2026-11-03T05:50:17.606Z","publishedAtSource":"pipeline","updatedAt":"2026-11-03T05:50:17.606Z"},"license":"Open for reading and citing with a link to https://findausecase.com/use-cases/cloudflare-runs-a-multi-agent-ai-code-reviewer-on-merge-requests-across-5-169-repositories. Bulk republication requires permission."}