Cloud Automation fails with Environment error after "Ran update script"

[2222]

Where does the bug appear (feature/product)?

Somewhere else…

Describe the Bug

Cloud Automation fails during environment setup with “Environment error” after “Ran update script”. Runs start after enabling on-demand spending, but fail before any agent work begins. Error message: “Agent encountered an error. We encountered an unexpected error repeatedly.” Expanding “Environment error” shows no useful detail.

Steps to Reproduce

1. Enable on-demand spending in Cursor Dashboard ($10/month cap).
2. Create a scheduled Automation targeting a private GitHub repo (branch: main).
3. Use minimal smoke test instructions: create one markdown file and git push.
4. Run the Automation manually (Run now).
5. Observe: status shows started/running, then fails with Environment error.

Also tried:

• Added .cursor/environment.json with install: “echo no-op” — no change
• Switched models (GPT-5.5 High → Composer 2.5) — no change
• No custom Team environment; Cloud dropdown only

Expected Behavior

Smoke test should complete: create a test markdown file, commit, and push to main. Automation log should show “smoke test done”.

Screenshots / Screen Recordings

Operating System

MacOS

Version Information

Version: 3.10.20
VS Code Extension API: 1.125.0
Commit: 23b9fb205fe595ea2be29da7214e19762d037fc0
Date: 2026-07-07T07:03:33.071Z
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: Darwin arm64 25.5.0

For AI issues: which model did you use?

GPT-5.5 High and Composer 2.5 — same Environment error with both

Additional Information

Private personal Obsidian vault on GitHub. Privacy Mode is active (cloud storage allowed), not legacy. On-demand spending was disabled initially — that caused “Agent failed to start”; enabling it fixed startup but execution still fails at environment setup. Persists across models. No vault contents involved — this is a technical infrastructure issue only.

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

[deanrie]

Hey, thanks for the detailed report. This looks like the same case we discussed in another thread recently, a crash during environment setup before the model even starts, so switching models won’t change anything.

Key point: deleting .cursor/environment.json from the repo does not clear the saved environment on our side. It stays there and keeps getting used. That’s why echo no-op didn’t help. There’s no separate Delete button for environments in the dashboard right now, so you need to clear it another way.

Try this in order:

1. Force a snapshot rebuild: make any change to environment.json or setup.sh and push it. This should invalidate the current snapshot and trigger a rebuild. Note: if a broken git state is already stuck in the base layer, changing the config might not rebuild from scratch. If that happens, go to step 2.
2. Reconnect the GitHub repo via Integrations, reconnect or re-authorize, or create a new environment on top of the same repo with a different name. For another user with the exact same issue, reconnecting the repo recreated the environment and fixed it. Details: Environment Error when running Cloud Agent. If the same repo is connected under a different Cursor account, that could also interfere, so it’s worth checking.
3. Run Automation again.

If it still doesn’t help, temporarily turn off Privacy Mode, run it again, and send a fresh Request ID bcId for that run. With that, we can look at the logs for that specific run and confirm the cause.

Let me know how it goes.