API quickstart

One endpoint to check risky inputs.

Send text, images, PDFs, documents, OCR output, or extracted content to POST /v1/scan. Mighty returns an allow, review, or block action before people, AI tools, OCR, or automation rely on the material.

Start with curlView fields
Fast test path/v1/scan
1
Send one payload

JSON text, multipart file, or raw binary.

2
Set workflow context

Mark the scan as input, output, or workflow-specific material.

3
Read the route

Use action, risk score, signals, and scan_id in your app.

allow
review
block
Copy this first

A test integration is one POST.

Put Mighty before the workflow step that reads customer material. Start with one file or text blob, then wire the returned action into your existing review path.

01
file
02
scan
03
route
Endpoint
/v1/scan
Payloads
JSON, multipart, binary
Result
allow, review, block
curl -X POST https://gateway.trymighty.ai/v1/scan \
  -H "X-API-Key: $MIGHTY_API_KEY" \
  -F "file=@claim-packet.pdf" \
  -F "scan_phase=input" \
  -F "mode=secure"
Use first

Run against one real intake flow before changing downstream automation.

Read first

The response includes action, risk score, signals, scan_id, and detected content type when available.

Authentication

Send API keys with X-API-Key. Bearer tokens using the mc_ key format are also accepted by the gateway.

Base URL for this environment
https://gateway.trymighty.ai

JSON text scan

Use this when the workflow already has text or OCR output.

curl -X POST https://gateway.trymighty.ai/v1/scan \
  -H "X-API-Key: $MIGHTY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "Text or OCR output to inspect",
    "content_type": "text",
    "scan_phase": "input",
    "mode": "secure",
    "profile": "balanced"
  }'

Multipart file scan

Use this for PDFs, images, and office documents from intake.

curl -X POST https://gateway.trymighty.ai/v1/scan \
  -H "X-API-Key: $MIGHTY_API_KEY" \
  -F "file=@claim-packet.pdf" \
  -F "scan_phase=input" \
  -F "content_type=pdf" \
  -F "mode=secure"

Raw binary scan

Use this when you stream an image or PDF body directly.

curl -X POST "https://gateway.trymighty.ai/v1/scan?content_type=image&scan_phase=input&mode=secure" \
  -H "X-API-Key: $MIGHTY_API_KEY" \
  -H "Content-Type: image/jpeg" \
  --data-binary "@damage-photo.jpg"

Request fields

JSON, multipart form data, and raw binary uploads all route through the same scan endpoint. Raw binary scans pass options as query parameters.

content
Required for JSON text scans. For multipart scans use file or content.
file
Multipart file field for PDFs, images, and documents.
scan_phase
Required. Use input for submitted material or output for generated/extracted output.
content_type
Optional. auto, text, image, pdf, or document. Defaults to auto.
mode
Optional. fast, secure, comprehensive, or auto. Defaults to secure.
analysis_mode
Legacy alias for mode. Accepted for older clients.
profile
Optional. Defaults to balanced. Supports profiles such as strict, balanced, permissive, code_assistant, and ai_safety.
data_sensitivity
Optional. standard, tolerant, or strict. Defaults to standard.
context
Optional workflow context string, for example insurance_claim_intake, repair_estimate_review, damage_photo_review, logistics_pod_review, construction_change_order, or ocr_output_review.
original_prompt
Optional original prompt when scanning model output.
metadata
Optional JSON object for your correlation metadata.
session_id
Optional stable session identifier.
scan_group_id
Required when scan_phase=output. Must be a UUID when supplied.
request_id
Optional UUID. Generated if omitted.
async
Optional boolean. Requires mode=comprehensive and image or pdf content.
webhook_url
Optional. Requires async=true.
ocr_engine
Optional. Current supported value is paddle.
stop_on_first_threat
Optional boolean for early stop behavior.
defer_enhance
Optional boolean. Supported with secure mode.

Response fields

Responses vary by content type, mode, and whether the scan is synchronous or asynchronous.

action
Recommended workflow action such as allow, review, or block.
risk_score
Numeric risk score returned by the scanner.
risk_level
Risk level label when included by the scanner.
threats
Detected categories or signals for the submitted material.
extracted_text
Text extracted from supported files when available.
content_type_detected
Detected input type such as text, image, PDF, or document.
scan_id
Identifier for correlating scan results and logs.
scan_group_id
Group identifier used to connect input and output scans.
scan_status
Status for asynchronous or staged scans when returned.
processing_ms
Processing time reported by the gateway when available.
page_results
Per-page document results when returned for PDFs/documents.
redacted_output
Redacted output when the scanner returns a sanitized version.