{
 "documentation_revision": 40,
 "api_revision": 86,
 "source": "Endpoints.all + Learn.endpoints + McpTools",
 "operation": {
  "summary": "Track a person's body for up to 2 minutes: the phone finds 33 body points in every camera frame (ML Kit, offline and free; no picture leaves the phone). Read the frames as they come with GET /v1/camera/pose/{id}",
  "method": "POST",
  "auth": true,
  "mcp_tool": "camera_pose_start",
  "query": [],
  "developer_only": false,
  "capture": true,
  "body": [
   {
    "name": "camera",
    "description": "front (default, the person can see the screen) or rear",
    "type": "string",
    "required": false,
    "enum": [
     "front",
     "rear"
    ]
   },
   {
    "name": "seconds",
    "description": "How long to track, 5-120 (default 60). It also stops when nobody has read it for 10 s",
    "type": "integer",
    "required": false
   },
   {
    "name": "preview_every_ms",
    "description": "How often to refresh the small preview picture, 150-2000 (default 400)",
    "type": "integer",
    "required": false
   },
   {
    "name": "preview_edge_px",
    "description": "Longest side of the preview picture, 120-480 (default 320)",
    "type": "integer",
    "required": false
   }
  ],
  "example": "POST /v1/camera/pose/start {\"camera\":\"front\",\"seconds\":45}",
  "idempotent": false,
  "path": "/v1/camera/pose/start",
  "capability": "camera",
  "returns": "session_id, state starting|running, camera, mirror (true for front: draw it mirrored), max_ms, started_at",
  "mcp_input_schema": {
   "type": "object",
   "properties": {
    "seconds": {
     "description": "How long to track, 5-120 (default 60). It also stops when nobody has read it for 10 s",
     "type": "integer"
    },
    "preview_edge_px": {
     "description": "Longest side of the preview picture, 120-480 (default 320)",
     "type": "integer"
    },
    "preview_every_ms": {
     "description": "How often to refresh the small preview picture, 150-2000 (default 400)",
     "type": "integer"
    },
    "camera": {
     "description": "front (default, the person can see the screen) or rear",
     "type": "string",
     "enum": [
      "front",
      "rear"
     ]
    }
   },
   "required": []
  },
  "errors": [
   "capability_disabled",
   "permission_denied",
   "notifications_required",
   "device_locked",
   "sensor_busy",
   "service_start_blocked",
   "invalid_parameter"
  ],
  "requires": null
 }
}