Skip to main content
The Files API stores organization-scoped documents in S3 and tracks their metadata in PostgreSQL. Use the presigned upload flow when you want clients, scripts, or CI jobs to upload files autonomously and trigger downstream classification.

​
Authentication

All endpoints require:
  • x-api-key
  • x-organization-id

Upload Files

Request presigned URLs, upload bytes directly to S3, then confirm the upload to trigger classification.

List File Readings

Retrieve extracted file readings after the async classification and processing steps complete.

​
Core Endpoints

​
Status Lifecycle

  • folder_id is optional on upload requests. Omit it to upload at the root level.
  • project_id is optional on PATCH /v1/files/batch-update. When present, the backend creates file_project links.
  • POST /v1/files/upload is still available, but the presigned flow is recommended for automation and large-file friendly clients.
  • POST /v1/files/send-event can be used to manually retry downstream processing when a classified file has not produced a file_reading yet.
  • Once processing finishes, clients can use GET /v1/files/readings, optionally edit extracted values with PUT /v1/files/readings/{file_reading_id}, and then create linked invoices or wastes with POST /v1/files/readings.
  • POST /v1/files/readings takes every reading of one file. Today a file has one reading, so that is one item.
  • Files and readings list their classifications in classifications (id, document_type, category, confidence_level, custom_label and a reading reference), next to the tags they always had. Every file has one classification today; the list is the groundwork for documents that bill several services, and no endpoint creates a second classification yet.

MCP Tools

Access file data from AI assistants via MCP

Invoices API

Manage invoices created from file readings