What problem does it solve? Sentry dashboards accumulate hundreds of unresolved issues, and deciding which ones deserve engineering attention is manual, error-prone work — made worse by the fact that Sentry's MCP search returns windowed counts that can under-report real impact by 30x, silently inverting triage decisions. ## Core Features & Use Cases - Gate-and-escalate triage: Pulls all not-resolved Sentry issues, applies a six-rule qualifying gate (status, staleness, severity, dedupe, parked, environment), and tags every non-qualifier with an explicit skip reason. - Count confirmation: Re-fetches true lifetime occurrence and user counts per issue before any threshold is applied, protecting against windowed MCP search results. - Cross-reference and escalation: Checks GitHub for existing linked issues via marker search, then files approved survivors through the github-issues filing script with preview-then-confirm and a burst rail above 5 candidates. - Use Case: Ask "what's blowing up in Sentry for project X" and receive a ranked report of qualifying production issues with confirmed lifetime counts, each either filed as a GitHub issue or annotated with why it was skipped. ## Quick Start Triage the top unresolved Sentry issues for my project and file GitHub issues for any that qualify.
