| # CleanCity Agent - MCP Server API Documentation |
|
|
| This document describes the **Model Context Protocol (MCP) tools** exposed by CleanCity Agent's MCP server. |
|
|
| ## 🔌 Server Information |
|
|
| - **Server Name:** `cleancity-agent` |
| - **Version:** 1.0.0 |
| - **Protocol:** MCP (Model Context Protocol) |
| - **Transport:** stdio |
|
|
| ## 🛠️ Available Tools |
|
|
| CleanCity Agent exposes **6 powerful tools** that AI agents can use to detect trash, plan cleanups, track events, and generate reports. |
|
|
| --- |
|
|
| ## 1. `detect_trash` |
| |
| Analyze an image and detect trash/litter using computer vision. |
| |
| ### Input Schema |
| |
| ```json |
| { |
| "image_path": "string (required)", |
| "location": "string (optional)", |
| "notes": "string (optional)" |
| } |
| ``` |
| |
| ### Parameters |
| |
| | Parameter | Type | Required | Description | |
| |-----------|------|----------|-------------| |
| | `image_path` | string | ✅ Yes | Absolute path to image file on disk | |
| | `location` | string | ❌ No | Location where trash was found (e.g., "Central Park") | |
| | `notes` | string | ❌ No | Additional context or observations | |
| |
| ### Returns |
| |
| ```json |
| { |
| "detections": [ |
| { |
| "object_type": "plastic_bottle", |
| "confidence": 0.95, |
| "bbox": [100, 150, 250, 400] |
| } |
| ], |
| "count": 5, |
| "annotated_image_path": "/path/to/annotated_image.jpg" |
| } |
| ``` |
| |
| ### Example Usage |
|
|
| **Claude Desktop:** |
| ``` |
| User: Analyze the trash in /home/user/beach_litter.jpg at Ocean Beach |
