LGM RAG Diagnostics Toolkit

Architect: Antigravity (Principal AI Architect) Project: timz-481219 (LGM Legal Auto)

Overview

This toolkit was created to diagnose and repair the "Missing Chunks" issue in the Vertex AI RAG system. It bypasses the standard CLI/Console limits using direct Python SDK calls.

Tools

1. repair_rag.py (Diagnosis)

• Purpose: Lists all Data Stores and reveals their hidden "Chunking Config".
• Usage: python repair_rag.py
• Key Output: Confirms if "Layout-Aware Chunking" is active.

2. import_titan.py (Ingestion)

• Purpose: Forces immediate ingestion of files from GCS into the Discovery Engine index.
• Usage: python import_titan.py
• Target: Currently set to ingest gs://titan-lgm-evidence-timz/TITAN_MEMORY/00_MASTER_PROTOCOLS.txt.

3. search_titan.py (Verification)

• Purpose: Runs a semantic search query against the live index to verify knowledge retrieval.
• Usage: python search_titan.py
• Query: "What is the total damages amount for non-compliance?"

4. list_docs.py (Audit)

• Purpose: Lists raw document IDs currently in the index.
• Usage: python list_docs.py

TITAN Protocols

The Master Protocols are located at:

• Local: c:\Workspace\LGM\TITAN_MEMORY\00_MASTER_PROTOCOLS.md
• Cloud (Active RAG): gs://titan-lgm-evidence-timz/TITAN_MEMORY/00_MASTER_PROTOCOLS.txt