Docs
Operations
How the work PulserNode does on your instance is run, bounded, and recorded.
1. Durable jobs
Builds, live tests, and repairs run as jobs on our workers, not in your browser. Closing the tab does not cancel them. When you come back, the page replays every step the job has recorded.
Each job carries your workspace, is leased to one worker at a time, and retries with backoff after a transient failure. A job is never run twice for the same request.
2. Limits every job runs under
| Limit | Value |
|---|---|
| Build deadline | 60 minutes |
| Repair attempts per failure | 3, then the circuit opens and a person is told |
| Deadline per repair attempt | 15 minutes |
| Repair workers per workflow | 1, enforced by a lock |
| New workflows per day on Basic | 3, resetting at 00:00 UTC |
3. What waits for you
- Plans are shown before anything is written to your instance.
- Changes that touch credentials, triggers, or remove nodes need explicit approval.
- Live tests and Guard replays run real executions, so they need your consent unless you turned on autonomous builds.
- Automatic repair only runs on workflows you enrolled, and can be switched off for the workspace.
4. What is kept
Workflow structure, version snapshots, and sanitized execution summaries are kept for the life of your workspace. Business payloads are not retained. Encrypted diagnostic samples expire after 30 days.
Disconnecting an instance removes our access immediately. The full retention schedule is in the privacy policy.
5. Status and incidents
The status page checks whether the PulserNode API responds. It does not report worker or MCP availability, or your n8n connection health. For an incident update, email hello@pulsernode.com with the reference shown on the error panel. Security reports go to security@pulsernode.com.