vss-query-analytics

To retrieve video analytics incidents and metrics from Elasticsearch, query the VA-MCP endpoint using a two-step JSON-RPC flow: initialize to obtain a session ID, then execute tools/call with curl and jq for data extraction.

Checking VSS alerts and sensor metrics without deploying a VLM is done by routing read-only queries through the VA-MCP server, allowing you to retrieve incidents and metrics across real-time or verification modes.

Yes, you can query per-sensor or multi-sensor analytics data by specifying the desired mode in your read-only VA-MCP request, retrieving incidents and metrics from the Elasticsearch-backed datasets for both real-time and verification contexts.

Yes, a reachable VA-MCP server on port 9901 is required to extract video analytics data, as the Skill relies on this endpoint to establish a JSON-RPC session and return read-only metrics, alerts, and incidents.

The main limitation is that querying video analytics metrics via the VA-MCP endpoint is strictly read-only, meaning you cannot deploy VLMs or generate narrative reports, and you are limited to extracting existing incidents and sensor data.

Yes, the VA-MCP query process requires curl and jq for data extraction, utilizing curl to execute the JSON-RPC initialize and tools/call requests and jq to parse the returned JSON metrics and incident responses.