MCP Image Recognition

---
name: mcp-image-recognition
category: osint-media-geolocation
cost: free
api_key_required: no
repo: https://github.com/mario-andreschak/mcp-image-recognition
paired_skills: ["reverse-image-search-workflow", "geolocation-from-image", "video-content-geolocation"]
capabilities: ["image-recognition", "vision-analysis", "geolocation"]
---

# MCP Image Recognition — vision-model describe/pivot

Wraps Anthropic or OpenAI vision APIs to extract text and visual elements from images. Used as a describe-then-search pivot for reverse image search and geolocation work.

## Install

```
npm i -g mcp-image-recognition
```

## Configuration

```json
{
  "mcpServers": {
    "image-recognition": {
      "command": "mcp-image-recognition",
      "env": {
        "ANTHROPIC_API_KEY": "YOUR_ANTHROPIC_KEY_HERE"
      }
    }
  }
}
```

Use `OPENAI_API_KEY` instead if you prefer the OpenAI vision backend.

## What it adds

Claude can read and describe an image's visual content mid-investigation — translating signs, classifying building styles, identifying vehicle types, reading text in a photo — then use those extracted elements to drive follow-on OSINT queries. Critical for the image-to-geolocation pipeline when the image itself has no EXIF data.

## Pairs with skills

- 011 `reverse-image-search-workflow`
- 013 `geolocation-from-image`
- 016 `video-content-geolocation`

## Cost

Uses Anthropic or OpenAI vision tokens. No additional cost beyond your existing Claude usage.