MCP Security Vision
by kfr0st
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CAMERAS | No | JSON for multiple cameras: object mapping camera IDs to sources, or array of camera objects. | |
| DB_PATH | No | SQLite database path. | data/security.db |
| MCP_HOST | No | HTTP transport host. | 127.0.0.1 |
| MCP_PORT | No | HTTP transport port. | 8000 |
| YOLO_CONF | No | Object detection confidence threshold. | 0.5 |
| FACE_MODEL | No | Face detection model: 'hog' or 'cnn'. | hog |
| FACE_SCALE | No | Scale at which faces are searched. | 0.25 |
| YOLO_IMGSZ | No | YOLO inference size; must be a multiple of 32. | |
| YOLO_MODEL | No | YOLOv8 model file. | yolov8n.pt |
| YOLO_DEVICE | No | Inference device: 'cpu', CUDA device index such as '0', or 'mps'. Default is auto. | |
| CAMERA_SOURCE | No | Camera source for a single camera: USB device index, RTSP URL, or 'synthetic'. | 0 |
| FACES_ENABLED | No | Enable or disable face recognition. | |
| MCP_TRANSPORT | No | MCP transport: 'stdio', 'sse', or 'streamable-http'. | stdio |
| SNAPSHOTS_DIR | No | Directory for generated snapshots. | snapshots |
| FACE_TOLERANCE | No | Maximum face distance to consider the same person. | 0.6 |
| RTSP_TRANSPORT | No | RTSP transport: 'tcp', 'udp', or 'auto'. | tcp |
| KNOWN_FACES_DIR | No | Directory containing known-face reference photos. | known_faces |
| ZONES_SEED_FILE | No | Initial restricted-zone seed file. | config/zones.example.json |
| ALERT_WEBHOOK_URL | No | JSON webhook URL for alerts. | |
| MCP_ALLOWED_HOSTS | No | Comma-separated allowed Host headers for HTTP transport. | |
| SNAPSHOT_ON_ALERT | No | Save JPEG evidence on each alert. | true |
| VISION_FAKE_MODELS | No | Use fake models for demo/CI without YOLO or dlib. | false |
| MCP_ALLOWED_ORIGINS | No | Comma-separated allowed Origin headers for HTTP transport. | |
| FACES_WITHOUT_PERSON | No | Search faces even without a person detection. | false |
| EVENT_COOLDOWN_SECONDS | No | Anti-spam cooldown for the same event. | 20 |
| FACE_STABILIZER_WINDOW | No | Face recognition stabilization window. | 3 |
| PROCESS_EVERY_N_FRAMES | No | Analyze AI every N frames. | 5 |
| RECONNECT_DELAY_SECONDS | No | RTSP reconnect backoff delay. | 5 |
| PRESENCE_TIMEOUT_SECONDS | No | Seconds without a detection before considering someone left a zone. | 2.5 |
| FACE_STABILIZER_MIN_VOTES | No | Minimum votes for face recognition stabilization. | 2 |
| INTRUSION_COOLDOWN_SECONDS | No | Anti-spam cooldown for zone re-entry. | 30 |
| MAX_RECONNECT_DELAY_SECONDS | No | Maximum RTSP reconnect backoff delay. | 60 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Server capabilities have not been inspected yet.
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
No tools | |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
This server cannot be deployed
Maintenance
ActivityMaintained
ResponsivenessNo issues