AI incident investigator
Production checks with evidence, not noise.
Buggy runs HTTP and real browser checks, captures screenshots, network failures and console errors, correlates them with deploys and commits, then asks an LLM for a conservative diagnosis.
HTTPstatus and body preview checks
ChromiumPlaywright browser evidence
LLMstructured diagnosis with fallback
buggy.check
$ run playwright checkout mode: real-browser · chromium target: /checkout network: GET /checkout 404 screenshot: /artifacts/failure.png hypothesis: Recent checkout refactor broke the route. Suggested fix: verify app/api/orders and route mapping.