Frigate Server
БесплатноНе проверенEnables AI assistants to interact with Frigate NVR security camera systems, supporting camera management, event detection, snapshots, recordings, and system sta
Описание
Enables AI assistants to interact with Frigate NVR security camera systems, supporting camera management, event detection, snapshots, recordings, and system stats via natural language.
README
A Model Context Protocol (MCP) server for Frigate NVR, enabling AI assistants to interact with your Frigate security camera system.
🌟 Features
- Camera Management: List and monitor all configured cameras
- Event Detection: Query detection events with filtering by camera, object type, and time
- Live Snapshots: Get current or historical camera snapshots
- Recordings: Access recording summaries and segments
- System Stats: Monitor Frigate performance, detector speed, and camera FPS
- Configuration: Retrieve complete Frigate configuration
📋 Prerequisites
- Python 3.10 or higher
- A running Frigate NVR instance
- Access to the Frigate HTTP API
🚀 Installation
1. Clone the repository
git clone https://github.com/yourusername/frigate-mcp.git
cd frigate-mcp
2. Create a virtual environment
python -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
3. Install dependencies
pip install -e .
4. Configure environment
Create a .env file in the project root:
# Required: URL of your Frigate instance
FRIGATE_FRIGATE_URL=http://localhost:5000
# Optional: API key if authentication is required
# FRIGATE_API_KEY=your_secret_key_here
# Optional: HTTP timeout in seconds (default: 30)
# FRIGATE_TIMEOUT=30
🔧 Configuration
The server uses environment variables for configuration. You can set them in:
.envfile (recommended for development)- Environment variables (recommended for production)
Configuration Options
| Variable | Description | Default | Required |
|---|---|---|---|
FRIGATE_FRIGATE_URL |
Base URL of Frigate instance | http://localhost:5000 |
No |
FRIGATE_API_KEY |
API key for authentication | None | No |
FRIGATE_TIMEOUT |
HTTP request timeout (seconds) | 30 |
No |
FRIGATE_SERVER_HOST |
Server host for SSE/HTTP modes | 0.0.0.0 |
No |
FRIGATE_SERVER_PORT |
Server port for SSE/HTTP modes | 8000 |
No |
Example Configurations
Local Frigate instance:
FRIGATE_FRIGATE_URL=http://localhost:5000
Remote Frigate with authentication:
FRIGATE_FRIGATE_URL=http://192.168.1.100:5000
FRIGATE_API_KEY=your_secret_key
FRIGATE_TIMEOUT=60
🎯 Usage
The server supports three operational modes:
1. STDIO Mode (Default)
For direct integration with MCP clients like Claude Desktop:
python -m frigate_mcp.server
# or
frigate-mcp
2. SSE Mode (Server-Sent Events)
For web-based clients with real-time updates:
frigate-mcp-sse
The server will start at http://localhost:8000/sse (configurable via environment variables).
Features:
- Real-time event streaming
- WebSocket-like experience over HTTP
- Easy to debug in browser dev tools
- Compatible with web applications
3. HTTP Mode (REST API)
For production deployments and API access:
STDIO Mode (for MCP clients):
python -m frigate_mcp.server
SSE Mode (for web clients):
frigate-mcp-sse
HTTP Mode (for REST API):
frigate-mcp-http
Integrating with MCP Clients
Claude Desktop (STDIO Mode)
Testing the Connection
Run the included test script to verify your Frigate connection:
python test_connection.py
Expected output:
============================================================
Frigate MCP Server - Connection Test
============================================================
🔧 Frigate Configuration:
URL: http://localhost:5000
API URL: http://localhost:5000/api
Timeout: 30s
📹 Testing /api/config (cameras)...
✅ Found 1 camera(s)
- front_door: enabled
📊 Testing /api/stats...
✅ Frigate version: 0.16.3
✅ Detectors: ['coral']
✅ Active cameras: ['front_door']
...
Running the MCP Server
Start the server in stdio mode (for MCP clients):
python -m frigate_mcp.server
Claude Desktop (STDIO Mode)
Add to your MCP client configuration (e.g., Claude Desktop config.json):
{
"mcpServers": {
"frigate": {
"command": "/path/to/frigate-mcp/.venv/bin/python",
"args": ["-m", "frigate_mcp.server"],
"env": {
"FRIGATE_FRIGATE_URL": "http://localhost:5000"
}
}
}
}
Web Clients (SSE Mode)
Connect to the SSE endpoint for real-time updates:
const eventSource = new EventSource('http://localhost:8000/sse');
eventSource.onmessage = (event) => {
const data = JSON.parse(event.data);
console.log('Received:', data);
};
REST API (HTTP Mode)
Make standard HTTP requests:
# Get cameras
curl http://localhost:8000/tools/get_cameras
# Get events
curl http://localhost:8000/tools/get_events?limit=5
# Get stats
curl http://localhost:8000/tools/get_stats
Or use any HTTP client:
import requests
response = requests.post(
'http://localhost:8000/tools/get_events',
json={'camera': 'front_door', 'limit': 10}
)
print(response.json()) "frigate": {
"command": "/path/to/frigate-mcp/.venv/bin/python",
"args": ["-m", "frigate_mcp.server"],
"env": {
"FRIGATE_FRIGATE_URL": "http://localhost:5000"
}
}
}
}
🛠️ Available Tools
The server provides 7 MCP tools for interacting with Frigate:
1. get_cameras()
List all configured cameras with their status and properties.
Returns:
[
{
"name": "front_door",
"enabled": true,
"width": 1920,
"height": 1080,
"fps": 5
}
]
2. get_events(camera, label, limit)
Get recent detection events with optional filtering.
Parameters:
camera(optional): Filter by camera namelabel(optional): Filter by object type (person,car,dog, etc.)limit(default: 10): Maximum events to return (1-100)
Example:
get_events(camera="front_door", label="person", limit=5)
Returns:
[
{
"id": "1234567890.123456-abcdef",
"camera": "front_door",
"label": "person",
"start_time": 1704067200.5,
"end_time": 1704067205.8,
"has_clip": true,
"has_snapshot": true,
"zones": ["entrance"],
"thumbnail": "http://localhost:5000/api/events/1234.../thumbnail.jpg"
}
]
3. get_stats()
Get Frigate system statistics and performance metrics.
Returns:
{
"service": {
"uptime": 86400,
"version": "0.16.3",
"storage": {...}
},
"detectors": {
"coral": {
"inference_speed": 8.5,
"detection_start": 1704067200.0
}
},
"cameras": {
"front_door": {
"camera_fps": 5.0,
"process_fps": 5.0,
"detection_fps": 1.2
}
}
}
4. get_event_details(event_id)
Get comprehensive details about a specific event.
Parameters:
event_id: Unique event identifier
Returns:
{
"id": "1234567890.123456-abcdef",
"camera": "front_door",
"label": "person",
"start_time": 1704067200.5,
"end_time": 1704067205.8,
"duration": 5.3,
"score": 0.87,
"zones": ["entrance"],
"has_clip": true,
"has_snapshot": true,
"media": {
"thumbnail": "http://localhost:5000/api/events/.../thumbnail.jpg",
"snapshot": "http://localhost:5000/api/events/.../snapshot.jpg",
"clip": "http://localhost:5000/api/events/.../clip.mp4"
}
}
5. get_snapshot(camera, timestamp)
Get a snapshot URL from a specific camera.
Parameters:
camera: Camera nametimestamp(optional): Unix timestamp for historical snapshot
Returns:
{
"camera": "front_door",
"timestamp": "latest",
"url": "http://localhost:5000/api/front_door/latest.jpg",
"description": "Snapshot from front_door (latest)"
}
6. get_recordings(camera, date)
Get recording information for a specific camera and date.
Parameters:
camera: Camera namedate(optional): Date inYYYY-MM-DDformat (defaults to today)
Returns:
{
"camera": "front_door",
"date": "2024-01-01",
"recordings_count": 24,
"total_duration": 86400,
"recordings": [
{
"day": "2024-01-01",
"hour": "10",
"duration": 3600,
"events": 5
}
]
}
7. get_config()
Get Frigate configuration summary.
Returns:
{
"cameras": ["front_door", "backyard", "garage"],
"detectors": ["coral"],
"mqtt": {
"enabled": true,
"host": "localhost"
},
"model": "path/to/model.tflite",
"version": "0.16.3"
}
🏗️ Project Structure
frigate-mcp/
├── src/
│ └── frigate_mcp/
│ ├── __init__.py # Package metadata
│ ├── config.py # Configuration management
│ └── server.py # MCP server & tools
├── test_connection.py # Connection test script
├── pyproject.toml # Project dependencies
├── .env # Environment variables (not in git)
└── README.md # This file
Made with ❤️ for the Frigate and MCP communities
Установка Frigate Server
У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.
▸ github.com/dedsxc/mcp-frigateFAQ
Frigate Server MCP бесплатный?
Да, Frigate Server MCP бесплатный — установка в пару кликов через Unyly без оплаты.
Нужен ли API-ключ для Frigate Server?
Нет, Frigate Server работает без API-ключей и переменных окружения.
Frigate Server — hosted или self-hosted?
Доступен hosted-вариант: Unyly запускает сервер в облаке, локальная установка не обязательна.
Как установить Frigate Server в Claude Desktop, Claude Code или Cursor?
Открой Frigate Server на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.
Похожие MCP
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
автор: modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
автор: xuzexin-hzCompare Frigate Server with
Не уверен что выбрать?
Найди свой стек за 60 секунд
Автор?
Embed-бейдж для README
Похожее
Все в категории ai
