Enterprise authentication succeeds but API/Chat/Agent return [unauthenticated] Error (Windows 3.12.30)

Describe the Bug

Cursor Version: 3.12.30 Stable
OS: Windows 11

Issue:
I can log into Cursor successfully with my Enterprise account, but AI requests fail.

Network Diagnostics:
✓ DNS
✓ SSL
✓ Authentication
✓ Marketplace
✓ Authentication UI
✓ Cursor Tab
✓ Agent Endpoint

Failed:
✗ API
✗ Ping
✗ Chat
✗ Agent

Developer Console:
[transport] Connect error in unary AI connect
ConnectError: [unauthenticated] Error

PowerShell:
nslookup api2.cursor.sh → Success
Test-NetConnection api2.cursor.sh -Port 443 → TcpTestSucceeded: True

Already tried:

  • Updated Cursor to 3.12.30
  • Logged out and back in
  • Reinstalled Cursor
  • Restarted Windows
  • Changed HTTP/2 to HTTP/1.1
  • Tested on another network

Request ID:
feee0e1c-2b8d-40fc-bb68-dbb8202b6c11

Steps to Reproduce

  1. Open Cursor IDE (Version 3.12.30).
  2. Log in with my Enterprise account ([email protected]).
  3. Open a workspace.
  4. Open Chat or Agent.
  5. Send any prompt.
  6. The request fails with “[unauthenticated] Error”.

Network Diagnostics consistently show:

  • PASS: DNS, SSL, Authentication, Marketplace, Authentication UI, Cursor Tab, Agent Endpoint
  • FAIL: API, Ping, Chat, Agent

Developer Console shows:
[transport] Connect error in unary AI connect
ConnectError: [unauthenticated] Error

Expected Behavior

After signing in with my Enterprise account, Chat and Agent should authenticate successfully and respond to prompts without returning “[unauthenticated] Error”.

Screenshots / Screen Recordings

Operating System

Windows 10/11

Version Information

Version: 3.12.30 (user setup)
VS Code Extension API: 1.128.0
Commit: 63a2996a10d9e476b6c28e951dd7691d9c0cf480
Date: 2026-07-21T22:50:03.568Z
Layout: IDE
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
Node.js: 24.15.0
V8: 14.4.258.32-electron.0
xterm.js: 6.1.0-beta.256
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

All models (Chat/Agent). The authentication error occurs before a model response is generated.

For AI issues: add Request ID with privacy disabled

feee0e1c-2b8d-40fc-bb68-dbb8202b6c11

Additional Information

This issue affects my Enterprise account.

The IDE login succeeds and Enterprise Plan is displayed correctly.

Network Diagnostics:

  • PASS: DNS, SSL, Authentication, Marketplace, Authentication UI, Cursor Tab, Agent Endpoint
  • FAIL: API, Ping, Chat, Agent

Developer Console:
[transport] Connect error in unary AI connect
ConnectError: [unauthenticated] Error

I have already:

  • Updated Cursor to 3.12.30
  • Reinstalled Cursor
  • Logged out and back in
  • Restarted Windows
  • Switched HTTP/2 to HTTP/1.1
  • Tested on another network (mobile hotspot)
  • Verified nslookup and HTTPS connectivity

The issue persists.

Does this stop you from using Cursor

Yes - Cursor is unusable

Hey, thanks for the detailed report with the Request ID and diagnostics. That really helps.

Good news, this isn’t a client or network issue, so reinstalling, forcing HTTP/1.1, or switching networks couldn’t have helped. Notice the signature in Network Diagnostics: only reachability checks pass (DNS, SSL, Authentication, Marketplace, Cursor Tab, Agent Endpoint), but the authenticated requests fail (API, Ping, Chat, Agent). This pattern means you can reach our servers fine, but your session is being invalidated on the server side, which is why you see [unauthenticated].

The cause is Enterprise account provisioning via SSO and SCIM. Your email is periodically removed from the org during identity provider directory sync because it isn’t assigned to the Cursor app in the correct SCIM directory group. Each SSO login temporarily adds you back to the team, but the next sync (about once an hour) removes you again and kills the session. That’s why you see the cycle where login works, then everything breaks after a while.

What to do: please contact your IT team or the Cursor org admin at Silicon Labs and ask them to assign [email protected] to the Cursor application and add it to the correct directory group in your identity provider. Until that’s done, every login will be revoked at the next sync.

There’s also a known gap on our side in the re-enroll logic. It’s on our radar, but I can’t share an exact ETA yet. The main and fastest unblock right now is fixing the IdP setup.

SCIM and directory sync docs for your admin: SCIM | Cursor Docs

Let me know if the issue still happens after the IdP assignment is fixed.