Skip to main content
All Cursor integration capabilities

​
About

Ship your Cursor team’s agent telemetry to Sazabi using Cursor’s OpenTelemetry Export. Cursor runs the export on its own servers, so one team admin configures it once and it covers every member of the team across the Cursor IDE, the cursor-agent CLI, and Cloud Agents. Nothing is installed on developer machines. Events arrive under service cursor with the event name as the log body, including api_request (one per model request, with cursor.model.name and input, output, and cache token counts), api_error, skill_activated, hook_execution_complete, and the Cloud Agent events. Every event carries cursor.conversation.id and cursor.surface, and cursor.user.email when Cursor knows the member. Cloud Agent commit events (cloud_agent_commit) carry the commit SHA as vcs.ref.head.revision and the repository as vcs.repository.url, so you can tie agent work to the deploys that ship it.
Cursor’s Conversation content family exports prompt, response, and tool call text. It is off by default. Leave it off unless your team has agreed to send that text to Sazabi.

​
Prerequisites

  • A Cursor Enterprise plan. OpenTelemetry Export is not available on other Cursor plans.
  • A Cursor team admin to add the destination in Cursor Team Settings
  • Your Sazabi intake URL, shown above — open the log source and copy the Intake URL from its setup screen (the public key is embedded in the URL)

​
Set up in the dashboard

1

​
Create a destination in Cursor

As a Cursor team admin, open Cursor Team Settings, go to OpenTelemetry Export, and click Create destination.
2

​
Enter the intake URL and auth header

  • Base URL: your intake URL, exactly as shown above. Do not add /v1; Cursor appends /v1/logs and /v1/metrics itself.
  • Header: Authorization with the value Bearer sazabi_public_<key>, where <key> is the first part of your intake URL’s hostname, before the first dot.
3

​
Test and enable the destination

Click Test connection, then Enable. Cursor starts exporting within about a minute.Turn off the destination’s metrics signal. Sazabi stores Cursor’s metric data points as plain log lines, not as metrics, and the api_request events already carry the token counts per request.

​
Set up with the CLI

You can also register the Cursor log source with the Sazabi CLI (installed and authenticated — see CLI reference). Registering the source mints the same intake URL the dashboard shows above (the public key is embedded in its hostname):
The command prints the intake URL for the new source — copy the whole URL and enter it in Cursor using the steps shown above. Run sazabi log-sources get <log-source-id> at any time to reprint the endpoint, or sazabi log-sources list to see every log source in the project.

​
Verify

Run one agent request in the Cursor IDE, the CLI, or a Cloud Agent. Open the Intake page in the Sazabi dashboard and confirm events from the Cursor source appear within a few minutes. Start a thread and search for api_request to find model requests with token counts, or for a cursor.conversation.id to follow one agent session.

​
Troubleshooting

Test connection fails — Confirm the base URL is the intake URL with no path and no trailing /v1, and that the header value is Bearer, one space, then sazabi_public_ followed by the first part of your intake URL’s hostname. No events after enabling — Cursor does not backfill. Only activity after you enabled the destination is exported, so run a new agent request and wait a minute. The same event appears twice — Cursor delivers logs at least once. Events with the same cursor.event.id are one event. No commit events — cloud_agent_commit covers Cloud Agents pushing to GitHub repositories only. Commits made in the IDE or CLI, merge commits, and commits on the default branch are not exported. Log lines with JSON metric bodies — The destination’s metrics signal is on. Sazabi stores each Cursor metric data point as a plain log line rather than as a metric. Turn the metrics signal off in Cursor; per-request token counts come from the api_request events.

​
Further reading

Intake

See per-source ingestion volume and health.

Log sources

Browse the full log source catalog.