I created a Cursor automation that uses a Webhook about two weeks ago that seemed to work fine.
In the last few days it seemed to fail, even after regenerating a new token.
I even tried creating a complete new automation and encountered the same issue.
This is the error I encountered:
"code": "unauthenticated",
"message": "[unauthenticated] Error",
"details": [
{
"type": "aiserver.v1.ErrorDetails",
"debug": {
"error": "ERROR_NOT_LOGGED_IN",
"details": {
"title": "Authentication error",
"detail": "If you are logged in, try logging out and back in.",
"isRetryable": false
},
"isExpected": true
},
"value": "**base64 value**"
}
]
}
Steps to Reproduce
Create an automation with a webhook.
Copy its URL and authorization header into Postman.
Send a request.
Webhook trigger endpoint returns 401 for all external HTTP requests. It stopped working today.
Steps to Reproduce
1.Create an automation with a Webhook trigger (e.g., “Cursor Agent: Runbook”)
2.Save the automation — Cursor generates a webhook URL and auth header
3.Copy the webhook URL and auth header from the UI (“Copy auth header” button)
Send a POST request using curl:
curl -s -X POST “https://api2.cursor.sh/automations/webhook/”
-H “Authorization: Bearer crsr_”
-H “Content-Type: application/json”
-d ‘{“event”:“jira.label_added”,“issueKey”:“JIRA-9999”,“summary”:“Test”}’
Expected Behavior
The webhook triggers an automation run and returns a 2xx response.
Operating System
MacOS
Version Information
This is cloud agent
Additional Information
Actual behavior: The endpoint returns 401 Unauthenticated
Merged the threads for hygiene purposes and have one thread for easier communication.
Thank you for the detailed reports. We’ve identified this as a service-side issue affecting webhook triggers (and other automation endpoints) broadly. This is not specific to any individual account or token.
@premik91, your finding that the token works against api.cursor.com/v0/me confirms the token is valid: the issue is in the automations webhook auth path itself.
Our team is investigating, and I’ll update this thread when the issue is resolved.
Following up here. We’ve confirmed that the webhook authentication issue that was causing widespread 401 errors has largely stabilized.
That said, we are still seeing intermittent occurrences of this issue. Our engineering team is actively investigating the underlying root cause to ensure a permanent fix.
@Artur_Magalhaes, the MCP tool issue you mentioned may have been related to this same backend auth problem. If you’re still seeing MCP tools unavailable, could you confirm whether it’s specific to cloud agents/automations or also affecting other surfaces?
@pjv, noted on the status page feedback. I’ll pass that along.
If anyone is still experiencing consistent 401 errors on webhook triggers as of today, please let me know and I can investigate further.
Thanks for the automation ID. Regarding the automations getting stuck on environment creation, we identified and fixed an issue that was causing automation environments to fail during setup, specifically for automations with MCP servers configured. A fix has been deployed.
However, if your automation had a run that got stuck during this window, it may still be blocking new runs from starting. Could you try the following:
Go to your automation’s runs page and check if there’s a run stuck in a “Running” state
If so, stop that stuck run
Then try triggering a new run
If that doesn’t work, deleting the automation and recreating it with the same configuration should clear the stuck state and allow runs to start fresh.
On the Slack workaround – noted. Calling the Slack API directly via CURL/python is a solid approach in the meantime!
Let me know how it goes or if you face any issues.