TabulaRAG

A fast-ingesting tabular data MCP RAG tool backed with cell citations.

Quick start (infra)

1. Build and start services:
   ./scripts/dev-up.sh
2. Verify backend:
   http://localhost:8000/health
curl http://localhost:8000/health/deps
3. Stop services:
   ./scripts/dev-down.sh
4. Stream logs:
   ./scripts/dev-logs.sh or ./scripts/dev-logs.sh backend

Services:

• Frontend: localhost:5173
• Backend: localhost:8000
• Postgres: localhost:5433
• Qdrant: localhost:6333

Ingestion (CSV/TSV)

Upload a table for ingestion:

curl -F "file=@/path/to/data.csv" \
  -F "dataset_name=my_table" \
  http://localhost:8000/ingest

Notes:

• UTF-8 CSV/TSV only.
• Delimiter is inferred from the filename (.csv → comma, .tsv → tab).
• Header row: omit has_header to auto-detect by comparing the first two rows; set has_header=true or has_header=false to force.

Connecting via External Tools

TabulaRAG exposes two endpoints for integration with AI assistants and tool runners:

Note: If your client is running outside the browser (e.g. inside Docker or a desktop app), replace localhost with your machine's local IP address. Run ipconfig (Windows) or ifconfig (Mac/Linux) to find it.