Arke
ReferenceSchemas

Chunk

Reference for the arke/chunk@v1 profile -- text fragments for RAG and semantic search.

Profile

arke/chunk@v1

Required Properties

PropertyTypeDescription
labelstringChunk identifier (e.g. "Chunk 0")
contentstringText content of the chunk
sequencenumberOrder within the parent file

Optional Properties

PropertyTypeDescription
embedding_statusstringpending / complete
token_countnumberNumber of tokens in the chunk

Relationships

PredicatePeer TypeRequiredDescription
parentfileYesSource file this chunk was extracted from

Usage

Chunks are created automatically by the chunking pipeline when files are processed. They enable:

  • Semantic search -- Each chunk is independently embedded and indexed in Pinecone
  • RAG -- Relevant chunks are retrieved and passed to LLMs as context
  • Granular discovery -- Find specific passages within large documents

On this page