Show HN: Dogfood Bench

A new project called Dogfood Bench tracks whether Anthropic's Claude Code, which Anthropic uses to build Claude Code itself, is closing its own GitHub issues faster, using a script that pulls every issue in the anthropics/claude-code repository from the GitHub API and buckets each closure as fixed, duplicate or not planned. The page reports fix rate, closures by kind, open backlog, share of issues fixed within 30 days and median days to fix over complete UTC weeks, marking model releases with vertical lines and refreshing daily. Its author cautions that the page is "100% vibe coded with Claude Code" and that a falling fix rate can reflect growing issue volume from more users rather than decline, calling it "a half serious benchmark.

Dogfood Bench Is Claude Code fixing its own issues faster? Anthropic builds Claude Code with Claude Code. If AI makes software engineers more productive, the effect should show up in the claude-code issue tracker https://github.com/anthropics/claude-code/issues . This page measures it. Caveat: this page is 100% vibe coded with Claude Code. If the numbers are wrong, that is also a data point. Verdict, last 26 weeks Loading data. Fix rate Closures by kind Open backlog Fixed within 30 days Median days to fix Method A script reads every issue in the repository from the GitHub API. It records when each issue opened, when it closed, and the close reason. Each closure goes into one bucket by its reason. Who closed the issue does not matter, because the fixes supposedly come from AI anyway. - Fixed - Closed as completed, and not labeled invalid. - Duplicate - Closed as a duplicate of another issue. - Not planned - Closed as not planned, or labeled invalid. Most of these are bot sweeps of stale issues. Only fixed issues count as evidence of work. The page uses complete UTC weeks only. Vertical lines mark model releases. The data refreshes daily. Last fetch: unknown. Limits: "completed" is the default reason when someone closes an issue on GitHub. Some issues close as completed with no fix, for example when the reporter gives up. Issue volume also grows with the user count, so a falling fix rate can mean popularity, not decline. This is a half serious benchmark.