Lineage API Reference

Product: v0.19.0 · Contract: OpenAPI · Spec ops: Ingestion cancel & fairness

REST API endpoints for document lineage, chunk provenance, entity traceability, and multimodal assets

Convert vs ingest: PDF-sourced documents link pdf_id at admission. Lineage reflects the Insert (KG) phase — chunk/entity provenance is stamped after convert completes and Insert runs. PDF row Completed with doc still extracting is expected mid-pipeline. See Ingestion cancel & fairness.

Base URL

http://localhost:8080/api/v1

Endpoints Overview

Method	Path	Description
GET	`/documents/{id}/lineage`	Complete document lineage tree
GET	`/documents/{id}/metadata`	Document metadata (flat KV)
GET	`/documents/{document_id}/assets/{asset_id}`	Multimodal PNG by stable asset id
GET	`/documents/{document_id}/mm-assets/{*asset_path}`	Multimodal asset by relative path
GET	`/chunks/{chunk_id}`	Chunk detail with entities
GET	`/chunks/{chunk_id}/lineage`	Chunk lineage with parent refs
GET	`/entities/{entity_id}/provenance`	Entity source traceability
GET	`/lineage/entities/{entity_name}`	Entity lineage (all source docs)
GET	`/lineage/documents/{document_id}`	Document graph lineage (entities+rels)

Document Lineage

`GET /api/v1/documents/{document_id}/lineage`

Returns the complete lineage tree for a document: all chunks with position data and all entities extracted from them.

Path Parameters

Parameter	Type	Description
`document_id`	string	Document ID

Response 200 OK

{
  "document_id": "abc123def456",
  "chunks": [
    {
      "chunk_id": "abc123def456-chunk-0",
      "chunk_index": 0,
      "content_preview": "EdgeQuake is an advanced RAG framework...",
      "tokens": 150,
      "start_line": 1,
      "end_line": 25,
      "entity_count": 5
    },
    {
      "chunk_id": "abc123def456-chunk-1",
      "chunk_index": 1,
      "content_preview": "The pipeline processes documents through...",
      "tokens": 142,
      "start_line": 26,
      "end_line": 50,
      "entity_count": 3
    }
  ],
  "entities": [
    {
      "entity_id": "EDGEQUAKE",
      "entity_name": "EDGEQUAKE",
      "entity_type": "TECHNOLOGY",
      "chunk_ids": ["abc123def456-chunk-0", "abc123def456-chunk-1"]
    }
  ]
}

Error Responses

Status	Description
404	Document not found
500	Internal server error

Document Metadata

`GET /api/v1/documents/{document_id}/metadata`

Returns all stored metadata for a document as a flat JSON object. This data is from the KV storage key {document_id}-metadata.

Path Parameters

Parameter	Type	Description
`document_id`	string	Document ID

Response 200 OK

{
  "id": "abc123def456",
  "title": "paper.pdf",
  "document_type": "pdf",
  "file_size_bytes": 1048576,
  "sha256_checksum": "a1b2c3d4e5f6...",
  "page_count": 12,
  "status": "completed",
  "created_at": "2025-01-15T10:30:00Z",
  "processed_at": "2025-01-15T10:31:00Z",
  "llm_model": "gpt-4.1-nano",
  "embedding_model": "text-embedding-3-small",
  "chunk_count": 8
}

Notes: Fields vary depending on document source (PDF vs. text/markdown). PDF documents include pdf_id and modality fields when vision convert persisted mm-assets. The response includes all metadata stored during ingestion.

Multimodal Assets (mm-assets)

Vision PDF convert can persist page/chart PNGs and rewrite markdown figure links to ![alt](assets/page-0001.png). Assets are served under the linked document_id (set after convert links PDF → document).

`GET /api/v1/documents/{document_id}/assets/{asset_id}`

Binary PNG (or other image) by stable asset_id — filename stem from relative path (e.g. page-0001 from assets/page-0001.png).

curl -o page-0001.png \
  "http://localhost:8080/api/v1/documents/{document_id}/assets/page-0001" \
  -H "X-Workspace-ID: workspace-uuid"

`GET /api/v1/documents/{document_id}/mm-assets/{*asset_path}`

Same bytes by relative path under the document mm-assets root (e.g. assets/page-0001.png).

curl -o page-0001.png \
  "http://localhost:8080/api/v1/documents/{document_id}/mm-assets/assets/page-0001.png" \
  -H "X-Workspace-ID: workspace-uuid"

Lineage implication: chunk content may reference assets/… URLs; provenance traces back to PDF page via pdf_id + asset path. See Lineage Tracking Architecture.

Chunk Detail

`GET /api/v1/chunks/{chunk_id}`

Returns detailed information about a specific chunk, including extracted entities and relationships.

Path Parameters

Parameter	Type	Description
`chunk_id`	string	Chunk ID (format: `{doc_id}-chunk-{N}`)

Response 200 OK

{
  "chunk_id": "abc123def456-chunk-0",
  "document_id": "abc123def456",
  "document_name": "paper.pdf",
  "content": "EdgeQuake is an advanced RAG framework implemented in Rust...",
  "index": 0,
  "char_range": {
    "start": 0,
    "end": 1024
  },
  "start_line": 1,
  "end_line": 25,
  "token_count": 150,
  "entities": [
    {
      "id": "EDGEQUAKE",
      "name": "EDGEQUAKE",
      "entity_type": "TECHNOLOGY",
      "description": "An advanced RAG framework"
    }
  ],
  "relationships": [
    {
      "source_name": "EDGEQUAKE",
      "target_name": "RUST",
      "relation_type": "implemented_in",
      "description": "EdgeQuake is implemented in Rust"
    }
  ],
  "extraction_metadata": null
}

Chunk Lineage

`GET /api/v1/chunks/{chunk_id}/lineage`

Returns a chunk’s complete lineage chain — parent document info, position data, model provenance, and entity/relationship summary — in a single call.

Path Parameters

Parameter	Type	Description
`chunk_id`	string	Chunk ID

Response 200 OK

{
  "chunk_id": "abc123def456-chunk-0",
  "document_id": "abc123def456",
  "document_name": "paper.pdf",
  "document_type": "pdf",
  "chunk_index": 0,
  "content_preview": "EdgeQuake is an advanced RAG framework...",
  "tokens": 150,
  "start_line": 1,
  "end_line": 25,
  "start_offset": 0,
  "end_offset": 1024,
  "llm_model": "gpt-4.1-nano",
  "embedding_model": "text-embedding-3-small",
  "embedding_dimension": 1536,
  "entity_names": ["EDGEQUAKE", "RUST", "RAG"],
  "entity_count": 3,
  "relationship_count": 2,
  "file_path": "/uploads/paper.pdf",
  "created_at": "2025-01-15T10:30:00Z"
}

Implements

F3: Every chunk contains parent_document_id and position info
F8: PDF → Document → Chunk → Entity chain is traceable

Entity Provenance

`GET /api/v1/entities/{entity_id}/provenance`

Returns the full provenance trail for an entity: all source documents and chunks where it was extracted.

Path Parameters

Parameter	Type	Description
`entity_id`	string	Entity ID (UPPERCASE with underscores)

Response 200 OK

{
  "entity_id": "EDGEQUAKE",
  "entity_name": "EDGEQUAKE",
  "entity_type": "TECHNOLOGY",
  "description": "An advanced RAG framework implemented in Rust",
  "sources": [
    {
      "document_id": "abc123def456",
      "document_name": "paper.pdf",
      "chunks": [
        {
          "chunk_id": "abc123def456-chunk-0",
          "start_line": null,
          "end_line": null,
          "source_text": null
        }
      ],
      "first_extracted_at": null
    }
  ],
  "total_extraction_count": 2,
  "related_entities": [
    {
      "entity_id": "RUST",
      "entity_name": "RUST",
      "relationship_type": "implemented_in",
      "shared_documents": 1
    }
  ]
}

Entity Lineage

`GET /api/v1/lineage/entities/{entity_name}`

Returns source documents and chunks for a named entity.

Path Parameters

Parameter	Type	Description
`entity_name`	string	Entity name

Response 200 OK

{
  "entity_name": "SARAH_CHEN",
  "entity_type": "PERSON",
  "source_count": 2,
  "source_documents": [
    {
      "document_id": "abc123",
      "chunk_ids": ["abc123-chunk-0", "abc123-chunk-3"],
      "line_ranges": []
    }
  ],
  "description_versions": []
}

Document Graph Lineage

`GET /api/v1/lineage/documents/{document_id}`

Returns all entities and relationships extracted from a document, with extraction statistics.

Path Parameters

Parameter	Type	Description
`document_id`	string	Document ID

Response 200 OK

{
  "document_id": "abc123def456",
  "chunk_count": 8,
  "extraction_stats": {
    "total_entities": 20,
    "unique_entities": 15,
    "total_relationships": 12,
    "unique_relationships": 10,
    "processing_time_ms": null
  },
  "entities": [
    {
      "name": "EDGEQUAKE",
      "entity_type": "TECHNOLOGY",
      "source_chunks": ["abc123def456-chunk-0"],
      "is_shared": false
    }
  ],
  "relationships": [
    {
      "source": "EDGEQUAKE",
      "target": "RUST",
      "keywords": "implemented_in",
      "source_chunks": ["abc123def456-chunk-0"]
    }
  ]
}

SDK Usage Examples

Rust

use edgequake_sdk::EdgeQuake;

let client = EdgeQuake::new("http://localhost:8080")?;

// Document lineage
let lineage = client.documents().get_lineage("abc123").await?;
println!("Chunks: {}", lineage.chunks.len());

// Document metadata
let meta = client.documents().get_metadata("abc123").await?;

// Chunk lineage
let chunk = client.chunks().get_lineage("abc123-chunk-0").await?;
println!("Lines {}-{}", chunk.start_line.unwrap_or(0), chunk.end_line.unwrap_or(0));

TypeScript

import { EdgeQuake } from "@edgequake/sdk";

const client = new EdgeQuake({ baseUrl: "http://localhost:8080" });

// Document lineage
const lineage = await client.documents.getLineage("abc123");
console.log(`${lineage.chunks.length} chunks`);

// Document metadata
const meta = await client.documents.getMetadata("abc123");

// Chunk lineage
const chunk = await client.chunks.getLineage("abc123-chunk-0");
console.log(`Lines ${chunk.start_line}-${chunk.end_line}`);

Python

from edgequake import EdgeQuake

client = EdgeQuake(base_url="http://localhost:8080")

# Document lineage
lineage = client.documents.get_lineage("abc123")
print(f"{len(lineage.chunks)} chunks")

# Document metadata
meta = client.documents.get_metadata("abc123")

# Chunk lineage
chunk = client.chunks.get_lineage("abc123-chunk-0")
print(f"Lines {chunk.start_line}-{chunk.end_line}")

Error Handling

All endpoints return standard error responses:

{
  "error": "NotFound",
  "message": "Document 'abc123' not found"
}

Status	Meaning
200	Success
404	Resource not found
500	Internal server error

OpenAPI Documentation

Interactive API docs: http://localhost:8080/swagger-ui/
Machine-readable SSOT: edgequake_webui/openapi/openapi.snapshot.json

All lineage and asset endpoints are tagged in OpenAPI (Lineage, Documents).