docs(graphrag-sdk): align the GraphRAG SDK pages with the SDK
Merged
NA
6d ago+46+46+01234567890123456789-28-28-01234567890123456789
Aligns the GraphRAG SDK pages with what the SDK does today:
- Overview covers table ingestion, LLM-judged deduplication in
finalize(), opt-in text-to-Cypher and table re-sync. - Quick start pins the FalkorDB image, states the 4.20+ server requirement, and no longer crashes on the embedding size.
- Guides table adds Structured Ingestion, Reliability and Grounding, Reducing LLM Hallucinations and Accuracy Benchmark, and corrects the Graph Schema and Storage descriptions.
- New FAQ entries: server version, duplicate entities, counting questions, and the
markdownextra. - Guide links point to docs.falkordb.com/graphrag.
- GenAI tools index and the AWS toolkit FAQ describe the SDK accurately.
Summary by CodeRabbit
- Documentation
- Expanded the GraphRAG SDK guide with details on document and table ingestion, hybrid retrieval, synchronization, duplicate handling, and optional text-to-Cypher.
- Updated setup guidance, including the minimum supported FalkorDB version and the example embedder dimensions.
- Added FAQ coverage for supported file formats, table mappings, server-version checks, and other SDK capabilities.
- Revised the toolkit FAQ to describe ontology discovery, table ingestion, and multi-turn conversation history.
- Updated the GraphRAG documentation link.