alexpovel/issuedigger

Leverages Cloudflare's vector database (Vectorize) and embeddings API to semantically index entire issue threads, treating comments and initial posts as unified units to surface contextually similar issues rather than just keyword matches. Operates as a GitHub App that auto-indexes new issues on installation, supports manual re-triggering via comments, and handles backfilling with rate-limit awareness. The serverless architecture uses Cloudflare Queues for async processing and Workers AI for embeddings, keeping operational costs minimal while remaining self-hostable via standard infrastructure-as-code configuration.