Replace the raw Ollama fetch-based OCR with the Vercel AI SDK,
adding support for OpenAI, Google Gemini, Anthropic, and Ollama
as selectable providers from the Settings page. Uses generateText
with Output.object() and Zod schemas for type-safe structured
data extraction.
Made-with: Cursor
Tests S3 bucket access and falls back to a raw HTTP fetch to show
what the endpoint actually returns (helps identify port misconfiguration).
Made-with: Cursor
- Add start.sh that runs prisma db push before starting the server,
creating tables automatically on first deploy
- Copy prisma CLI + engines into production image for runtime migrations
- Add configurable storage cleanup: source PDF retention (default 30d)
and image retention (default 180d) with manual trigger in Settings
- Add /api/cleanup endpoint (GET for status, POST to run)
- Add retention settings to AppSettings schema
- Add ListObjectsV2 helper to minio.ts for bulk cleanup
Tested: local Docker build passes
Made-with: Cursor
