> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pnbr.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Prepare extraction

> Returns a ready-to-run `/v1/extractions/shape` call for this document. Extraction itself is the governed command surface.



## OpenAPI

````yaml /api-reference/openapi.json post /v1/documents/{id}/extract
openapi: 3.1.0
info:
  title: Penumbra Platform API
  version: 1.0.0
  description: >-
    The Penumbra /v1 REST surface. Developer-intent nouns (memories, documents,
    submissions, connections, preflight, bridge) over a typed, governed
    knowledge graph. Every write stages through a reviewable delta; verdicts are
    dispositions and findings, never scores.
servers:
  - url: https://pnbr.io
security:
  - bearerAuth: []
tags:
  - name: Memories
    description: Store and recall agent memory as first-class REST.
  - name: Documents
    description: Add and search documents over the source/file substrate.
  - name: Submissions
    description: A generic capture door for app events, forms, and payloads.
  - name: Search
    description: Hybrid, graph, and retrieval search.
  - name: Ontology & Shapes
    description: Read the active ontology, types, and shapes.
  - name: Deltas
    description: 'The governed-write primitive: stage, plan, apply, revert.'
  - name: Graph
    description: >-
      Direct reads of the typed graph: nodes, edges, and counts. Scope:
      graph:read.
  - name: Extraction
    description: Coerce source material into typed entities through a shape.
  - name: Context & Slices
    description: >-
      Author reusable context scopes (slices) and compile them into context. A
      slice is a saved scope; the context verbs act on a scope or a saved slice.
  - name: Preflight
    description: >-
      Read-only decision-quality verdicts: is the graph fit to act on?
      Disposition plus findings, never a score.
  - name: Loops
    description: Bounded agentic extraction and hydration loops over your context.
  - name: Connections
    description: Headless integration connect for token-paste platforms.
  - name: Bridge
    description: Map an external API's type system onto your ontology.
paths:
  /v1/documents/{id}/extract:
    post:
      tags:
        - Documents
      summary: Prepare extraction
      description: >-
        Returns a ready-to-run `/v1/extractions/shape` call for this document.
        Extraction itself is the governed command surface.
      operationId: extractDocument
      parameters:
        - name: id
          in: path
          required: true
          schema:
            type: string
      requestBody:
        content:
          application/json:
            schema:
              type: object
              properties:
                shape_id:
                  type: string
                project_id:
                  type: string
            example:
              project_id: 9142fa4d-7aad-4a1b-8f99-b522fd37518d
      responses:
        '202':
          description: Extraction handoff body.
          content:
            application/json:
              example:
                status: ready
                document_id: 9784c0d0-13dc-4704-91a5-e32afc953e06
                next_action:
                  method: POST
                  path: /v1/extractions/shape
                  body:
                    project_id: 9142fa4d-7aad-4a1b-8f99-b522fd37518d
                    source_file_id: 9784c0d0-13dc-4704-91a5-e32afc953e06
                request_id: ebde3ee7-bbe9-4a9d-aa95-6bec84b8b84f
components:
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: >-
        A `pnbr-` API key. Project-scoped keys make `project_id` implicit;
        org-scoped keys require it.

````