Unique identifier for the chunk (typically from the vector store).
The text content of the chunk.
Section heading or title the chunk belongs to, if available.
File path or document source path this chunk was extracted from.
Relevance score (0 to 1) indicating how well this chunk matches the query. Higher is better.
Which retrieval method produced this chunk.
'vector' — Dense vector similarity search'graph' — Knowledge graph traversal (GraphRAG)'research' — Iterative deep research synthesis
A single chunk of content retrieved during the retrieval phase.