video-analytics

You retrieve video analytics incidents by querying Elasticsearch through the VA-MCP JSON-RPC server. This requires a two-step MCP workflow: initializing a session ID header, then sending a tools/call request with the appropriate video_analytics_* method and parameters.

Yes, you can compute average speeds and people counts over a specific time range using video sensor signals. The Skill applies analytics methods to calculate occupancy-like statistics and summaries across your specified sensors, places, and time windows.

To query VA-MCP for recorded safety violations and object metrics, you need a running VA-MCP JSON-RPC server at port 9901. You must also know your target sensor IDs and places to correctly scope the video_analytics_* method parameters.

The two-step MCP workflow functions by first sending an initialization request to obtain a session ID header. After initializing the session, you send a tools/call request containing the video_analytics_* method and your query parameters to retrieve the video analytics data.

When drilling into a specific incident's details via the VA-MCP server, the limitation is that optional data like object IDs and extra info depends on what was originally recorded in Elasticsearch. You must specify these optional parameters in your tools/call request to retrieve the extended information.