Workspace usage
What your workspace's usage meters count, what they never count and how to read them.
Updated 09/10/2026
Your workspace keeps a running picture of the work Nexra does for you, so you can see how your integrations and agents are being used. For plans, see pricing or talk to us.
Governed executions
A governed execution is one write that ran successfully through Nexra's governed-action gate, whether a policy allowed it or a person approved it. Writes from flows, dead-letter retries, agents over MCP and the console gateway all count the same way.
Each action counts once. A flow record, its dead-letter retries and an agent's retries under the same idempotency key share one action, so a retry that succeeds after an earlier attempt failed is still one execution.
These writes never count:
writes that were denied or refused, including R4 writes;
writes still waiting for approval, and writes that were rejected;
writes that failed at the provider;
repeated dispatches refused as duplicates.
Never counted
Everything that helps you understand, test and prove your integrations is left out of your usage:
Reads, on every connected system. Reads are never blocked by usage.
Discovery: the connector catalogue, MCP tool discovery and capability search.
Policy tests: "Test this policy" and decision explain.
Connector validation runs.
Checks runs and release gate evaluation.
Nexra Agent lookups.
Connected systems
A connected system is a connection that is active, in the production environment, with a current credential held in the secure vault. Sandbox, demo and sample connections never count, so you can build and test as much as you like.
Connected systems are measured as the peak number in each calendar month (UTC). Nexra records the count every hour, after each counted write and whenever usage is read.
Your usage
Members with the usage.read permission can read the current month's meters:
GET /api/console/billing/usage{
"workspaceId": "<workspace id>",
"period": {
"key": "2026-10",
"start": "2026-10-01T00:00:00.000Z",
"end": "2026-11-01T00:00:00.000Z",
"elapsedFraction": 0.2667
},
"meters": [
{ "key": "connect.governed_executions", "measure": "count", "used": 1284, … },
{ "key": "connect.connected_systems", "measure": "peak", "used": 6, "current": 6, … }
],
"notMetered": [
"reads",
"explain",
"validation_runs",
"checks_runs",
"agent_read_tools",
"mcp_discovery"
],
"generatedAt": "2026-10-09T09:20:00.000Z"
}Each meter also reports a projection for the month.
Privacy
Usage records hold IDs only (the action, connection and run), never payloads. Usage is kept separately from the evidence chain, so usage never changes your evidence.
Agent units
Nexra Agent answers and model-judged checks draw on your workspace's agent units, shown in the Nexra Agent side panel. They are separate from governed executions.