Your n8n instances.
One place to check.

1. Create a Rewind monitoring key.

Sign in, choose the correct client workspace, then open n8n Monitor → Connect instance. Name the key after the instance, such as “Acme production”. Each key identifies one instance; use separate keys for separate instances. Viewers can read summaries; owners and operators can connect instances.

2. Keep the n8n key on your machine.

In your own n8n instance, open Settings → n8n API. Create an API key. Where scoped keys are available, select only workflow:list and execution:list. Non-Enterprise n8n keys have broader permissions even though this connector only uses GET requests. Keep that key private and revoke it when no longer needed.

The public API is available on self-hosted editions and supported paid n8n Cloud plans, not the Cloud free trial. Rewind does not host n8n. n8n API documentation ↗

Use Node.js 24 or later on a machine that can reach n8n. Load the keys as environment variables through your private environment or secret manager. Do not put keys in command arguments, Git, this JSON file, or chat.

• N8N_PRODUCTION_KEY: your n8n API key.
• REWIND_PRODUCTION_MONITOR_KEY: the Rewind monitoring key from step 1.

3. Describe the instance.

Save this as monitor.json. Replace the URL with your n8n base URL, without /api/v1. HTTPS is required except for local loopback HTTP.

{
  "instances": [
    {
      "url": "https://your-instance.example.com",
      "apiKeyEnv": "N8N_PRODUCTION_KEY",
      "rewindKeyEnv": "REWIND_PRODUCTION_MONITOR_KEY"
    }
  ]
}

Add another entry with separate environment-variable names for each instance. One process supports up to five instances; each can report to a different client workspace through its Rewind key.

4. Sync once, then keep it running.

npx --yes https://rewind.kanishq.dev/integrations/rewind-cli-0.2.0.tgz monitor --config monitor.json --once

Check that the correct instance appears in Rewind. Then start polling:

npx --yes https://rewind.kanishq.dev/integrations/rewind-cli-0.2.0.tgz monitor --config monitor.json

The package is downloaded directly from Rewind, not an npm registry listing. It has no install scripts or runtime dependencies. The default interval is three minutes; --interval 600 changes it to ten minutes. Keep the process running on your own machine or server. Closing it stops updates. Ctrl+C stops polling without changing any n8n workflow.

The dashboard refreshes its stored view every minute while visible. Refreshing the dashboard does not trigger a new n8n poll. “Sync overdue” means the last report is older than ten minutes or 2.5 times the configured interval, whichever is longer.

What gets sent.

The connector sends your instance URL, workflow IDs and names, active flags and update times, and execution IDs, workflow IDs, statuses and start/stop times. It drops node definitions, pinned data, credentials, execution input/output and raw errors before upload. Workflow names and URLs can still be sensitive; use this only for instances you are authorized to monitor.

n8n may return workflow definitions to the local connector as part of its workflow-list API. These are discarded locally. The n8n key never goes to Rewind. Rewind stores only the latest successful summary and the latest sync status per instance. A failed poll keeps the last good snapshot with a visible error.

Workspace members can read its summaries. Disconnecting removes the stored summary and revokes that Rewind key; it does not revoke the n8n key. Account/workspace exports include monitoring data, and account deletion removes owned monitoring data. Private infrastructure backups have separate retention. Full data handling →

Current limits and honest boundaries.

Five instances per workspace, sharing your existing ten-key account allowance. Each snapshot holds up to 200 workflows, 100 recent executions and 50 retained error executions; a partial-result message appears when there are more. The service has a shared 100-instance beta limit. API-key visibility, n8n retention, unsaved runs and activity between polls limit what can be shown. This is not a complete execution archive or an uptime guarantee.

Failures are records of previous runs, not unresolved incidents. There are no email/Slack alerts, automatic retries, workflow editing, activation or hosting in this release. Open the linked execution in n8n for its details. Monitoring does not record before-state or make arbitrary workflows recoverable. Configure supported recovery separately →

The beta remains free with no subscription. Running your own n8n and connector may use resources you already pay for. Your provider costs are separate.

If the first sync fails.

Authentication: check both key scopes and expiry; n8n Cloud trials do not expose this API. Unreachable: check the base URL and connector network. Redirects are refused to protect keys; use the final HTTPS URL. Unsupported response: check n8n compatibility or unusually large workflows. Capacity: disconnect an unused instance. Clock: keep the connector’s clock synchronized.

To rotate the Rewind key, stop the connector, disconnect the old instance, create a new monitoring key, update its environment variable, and restart. The previous summary is removed on disconnect. Rotating the n8n key only requires updating its environment variable and restarting.