Repair the RAG Vector Contract

Fix an index/query embedding dimension mismatch without corrupting vectors.

A intermediate AI Engineering challenge worth 10 points. Solve it hands-on in a real AI Engineering environment in your browser - no local setup, no fake shells.

The Challenge

Scenario: The runbook retriever crashes because query vectors are configured for 768 dimensions while the MiniLM index contains 384-dimensional vectors.

Repair pipeline.json so indexing and querying use the exact same approved embedding contract. Do not pad or truncate vectors. Run python diagnose.py, inspect the ranked IDs, then run check.