Analyze Document

apiVersion: naftiko.io/v1
kind: Capability
metadata:
  name: analyze-document
  version: 1.0.0-alpha2
  labels:
    provider: parseflow
    family: analyze
    domain: document-ai
spec:
  title: Analyze Document
  description: |
    Run a lightweight analysis pass over a document or text payload to surface
    chunk statistics, extraction candidates, and quality signals without
    committing the document to the search index.
  sharedAPI:
    $ref: ./shared/parseflow-api.yaml
  inputs:
    - name: file
      type: file
    - name: text
      type: string
    - name: mode
      type: string
      enum: [deterministic, byok_assisted]
      default: deterministic
    - name: chunk_size
      type: integer
      default: 2000
    - name: overlap
      type: integer
      default: 200
    - name: preset
      type: string
    - name: byok_provider
      type: string
      enum: [openai, anthropic, google]
    - name: byok_model
      type: string
  consume:
    http:
      method: POST
      path: /v2/analyze
      contentType: multipart/form-data
  expose:
    rest:
      path: /capabilities/analyze-document
      method: POST
    mcp:
      tool: parseflow_analyze_document
      description: Inspect a document and return chunk and extraction analysis.
  outputs:
    contentType: application/json