Importer Service - API Reference

Original filename

File content in bytes

URL to be cited in inline citations

The job identifier to query

Whether the job was found

Error message if job not found

The job status details (if found)

Direct file content

URL to download the file from

Original filename

Whether to store in S3

Batch size for importing documents, entities, and relationships

Whether to enable chunk embeddings

Whether to enable edge embeddings

Maximum tokens per chunk

Overlap tokens between chunks

Entity types to extract

Relationship types to extract

Partition id

Enable semantic units (web URLs; extract embedded PDF images + FileManager upload when true)

Generate multimodal descriptions (requires enable_semantic_units)

Minimum tokens per chunk (optional)

Custom separators for chunking (optional)

Whether to preserve separator in chunks (optional)

Smart Graph Attribute

Must be 1 for new SmartGraph or sharded enterprise graph (IVF index per-shard; empty shards -> ERR 9). Other values rejected.

Whether the graphs must be disjoint

An array of collection names that is used to create Satellite Collections

Enable strict filtering of entities and relationships based on provided types

default is 1

Number of lists for approximate search. If not set, computed as 8 * sqrt(collection_size)

Distance metric for vector index (default: "cosine")

Whether to use HNSW index (defaults to False)

Whether to enable community embeddings (default true when unset)

If true, chunk by separators only (ignores token size limits)

Dictionary of custom prompts (key: prompt name, value: prompt text). If non-empty, custom prompts will be used instead of defaults.

"vector_rag" or "full_graphrag" (empty defaults to full_graphrag)

PATH B: RAG file ID from File Manager. When set, source (file_content/file_url) is ignored and the file is fetched by ID.

Whether to enable semantic unit embeddings

List of files to process

Whether to store in S3

Batch size for importing documents, entities, and relationships

Whether to enable chunk embeddings

Whether to enable edge embeddings

Maximum tokens per chunk

Overlap tokens between chunks

Entity types to extract

Relationship types to extract

Partition id

Enable semantic units (web URLs; extract embedded PDF images + FileManager upload when true)

Generate multimodal descriptions (requires enable_semantic_units)

Minimum tokens per chunk (optional)

Custom separators for chunking (optional)

Whether to preserve separator in chunks (optional)

Smart Graph Attribute

Must be 1 for new SmartGraph or sharded enterprise graph (IVF index per-shard; empty shards -> ERR 9). Other values rejected.

Whether the graphs must be disjoint

An array of collection names that is used to create Satellite Collections

Enable strict filtering of entities and relationships based on provided types

Maximum number of extraction iterations to refine entity extraction (default is 1)

Number of lists for approximate search. If not set, computed as 8 * sqrt(collection_size)

Distance metric for vector index (default: "cosine")

Whether to use HNSW index (defaults to False)

Whether to enable community embeddings (default true when unset)

If true, chunk by separators only (ignores token size limits)

Dictionary of custom prompts (key: prompt name, value: prompt text). If non-empty, custom prompts will be used instead of defaults.

"vector_rag" or "full_graphrag" (empty defaults to full_graphrag)

PATH B: RAG file IDs from File Manager. When non-empty, files are fetched by ID and files (field 1) are ignored.

Whether to enable semantic unit embeddings

Whether files were processed successfully

Unique job identifier for tracking progress

Success or error message

Detailed error message if success is false

Service/pod identifier for orchestrator job→pod mapping (same as GenAI metadata)

Type of progress update

Progress message

Current file being processed (0-based)

Total number of files

Whether the operation succeeded

Error message if success is false

Unique job identifier

ISO 8601 timestamp of job creation

List of file names being processed

Number of files in the job

Current status of the job

Whether the job has finished (completed or failed)

Chronological history of status updates (oldest first)

Unique job identifier

ISO 8601 timestamp of job creation

List of file names being processed

Number of files in the job

Current status name

Whether the job has finished

Maximum number of jobs to return (default: 50)

Whether the request succeeded

List of job summaries (most recent first)

The rag_partition_id to recluster

Whether the recluster was accepted

Job id for GET /v1/jobs/{job_id}

Success or error message

Service/pod identifier

Status name (e.g., "GRAPH_BUILDER_STARTED")

ISO 8601 timestamp of this status update

Progress percentage (0.0 to 100.0)

Optional message with additional details