Pattern: Viral content triage — trigger → reverse search → geolocation analysis → assessment report

Skills used: media-verification-specialist, geolocation-verification-specialist, ai-document-analysis

Workflow overview

[Webhook Trigger] → [Compose Verification Plan] → [Reverse Image Search]
    → [Metadata Extraction] → [Geolocation Analysis] → [Assessment Report]

Node chain

1. Webhook Trigger

• Type: Webhook (POST)
• Path: /webhook/media-verification
• Body schema:

  {
    "media_url": "https://example.com/image.jpg",
    "claim": "Photo shows protest at City Hall on May 10, 2026",
    "source_post_url": "https://twitter.com/user/status/123456"
  }

2. Compose Verification Plan

• Type: HTTP Request
• Method: POST
• URL: {{ $vars.PATRIOT_API_URL }}/investigate/workflow
• Body:

  {
    "question": "Verify the authenticity and claimed location/time of this media: {{ $json.claim }}",
    "context": "Source URL: {{ $json.source_post_url }}"
  }

• Purpose: Get a structured verification plan from Claude.

3. Reverse Image Search

• Type: Code node (JavaScript)
• Purpose: Check the image against known reverse image search services:
• Google Lens (via URL parameter)
• TinEye API (if configured)
• Yandex Images (via URL parameter)
• Output: Earliest known appearance date, alternative sources, and any modifications detected.
• Note: Fully automated reverse image search may require commercial API access (TinEye). Manual steps can be flagged for analyst action.

4. Metadata Extraction

• Type: Code node or HTTP Request to Tika
• Purpose: Extract EXIF data from the image:
• Camera model, timestamp, GPS coordinates (if present)
• Software modification history
• Compression artifacts suggesting re-saves
• Note: Many social media platforms strip EXIF data. Absence of metadata is expected and should not be interpreted as evidence of manipulation.

5. Geolocation Analysis

• Type: HTTP Request (to Claude chat for analysis)
• Method: POST
• URL: {{ $vars.PATRIOT_API_URL }}/chat/message
• Body:

  {
    "message": "Analyze this image for geolocation clues. Identify visible landmarks, signage, vegetation, architecture, sun position/shadows, and any other location indicators. Provide a confidence assessment: exact location / city-level / region-level / insufficient evidence. Image URL: {{ $json.media_url }}",
    "persona": "investigator"
  }

• Output: Location assessment with confidence level and supporting evidence.

6. Assessment Report

• Type: Respond to Webhook
• Body:

  {
    "claim": "original claim text",
    "verification_status": "verified | disputed | inconclusive",
    "location_assessment": {
      "confidence": "city-level",
      "identified_location": "Atlanta, GA",
      "evidence": ["visible CNN Center signage", "Peachtree Street intersection"],
      "provenance": "AI-inferred"
    },
    "temporal_assessment": {
      "confidence": "date-level",
      "evidence": ["shadow angle consistent with May afternoon"],
      "provenance": "AI-inferred (Estimated)"
    },
    "earliest_known_post": "URL and timestamp from reverse search",
    "metadata_findings": "EXIF summary or 'metadata stripped'",
    "recommended_next_steps": ["field verification", "contact original poster"],
    "provenance_notice": "Location and temporal assessments are AI-inferred and require field verification"
  }

Confidence levels

When to use

• Verifying viral social media images or videos
• Fact-checking claims about protest locations or sizes
• Investigating potentially manipulated media
• Assessing the provenance of leaked documents or screenshots