Document Ingestion API

Import documents and convert them to searchable memories.

Overview

Document Ingestion processes various file formats and extracts content as memories:

• PDFs - Text extraction with optional OCR
• Images - Visual content analysis
• Audio - Automatic transcription
• URLs - Web page content extraction

Ingest File

Upload and process a file.

POST /v1/ingest/file

Form Data

Field	Type	Required	Description
file	file	Yes	File to ingest
chunk_size	int	No	Tokens per chunk (default: 500)
overlap	int	No	Chunk overlap (default: 50)
space_id	string	No	Store in this space
ocr	bool	No	Enable OCR (default: true)

Supported Formats

• Documents: PDF
• Images: PNG, JPG, GIF, WebP
• Audio: MP3, WAV, M4A
• Text: TXT, MD, CSV

Example

curl -X POST https://memoryapi.tensorheart.com/v1/ingest/file \
  -H "Authorization: Bearer $API_KEY" \
  -F "file=@document.pdf" \
  -F "space_id=documents" \
  -F "chunk_size=500"

Response

{
  "success": true,
  "data": {
    "document_id": "doc_abc123",
    "status": "completed",
    "memories_created": 15,
    "chunks_created": 15,
    "tokens_extracted": 7500
  }
}

---

Ingest URL

Extract content from a web page.

POST /v1/ingest/url

Query Parameters

Parameter	Type	Required	Description
url	string	Yes	URL to ingest
chunk_size	int	No	Tokens per chunk (50-5000, default: 500)
overlap	int	No	Overlap between chunks (0-1000, default: 50)
space_id	string	No	Store in this space

Security

Only public URLs are allowed. Private IPs, localhost, and cloud metadata endpoints are blocked.

Example

curl -X POST "https://memoryapi.tensorheart.com/v1/ingest/url?url=https://example.com/article&space_id=research" \
  -H "Authorization: Bearer $API_KEY"

---

Get Document Status

GET /v1/ingest/{document_id}

Example

curl https://memoryapi.tensorheart.com/v1/ingest/doc_abc123 \
  -H "Authorization: Bearer $API_KEY"

---

List Documents

GET /v1/ingest

Lists all ingested documents with their processing status.

Processing Pipeline

Upload → Extract Text → Chunk Content → Create Memories → Index for Search

Each chunk becomes a searchable memory, making your documents instantly queryable.