-
Notifications
You must be signed in to change notification settings - Fork 137
Description
Environment:
(Hint: "Report Extension Issue on Github" command will fill these out for you.)
Multiple environments tried:
- Windows 11
- Fedora Linux (latest Bazzite)
- Ubuntu 20.04
- Ubuntu 22.10
Version information
Cloud Code Extension version: Any from last year. Tried exclicitly:
- 2.68.0
- 2.51.0
- 2.48.0
- 2.23.0
VSCode version: 1.108.2
OS: Windows 11, Fedora Linux, Ubuntu (not host related)
Cloud SDK: N/A
Skaffold: N/A
Kubectl: N/A
Description: Paid Gemini AI Pro (Individual) subscriber. I am completely unable to use the extension because the authentication logic incorrectly identifies my personal account as an Enterprise/Cloud User.
There is no flow to select my personal license, or confirm that I am (not) an Enterprise User. I am locked into Entreprise Mode in this Extension on a Personal Account, with no way to enforce my (Paid) license to the Individual Pro,
The extension detects a "Shadow Project" (ID: quick-aileron-26f3p) on my Google account—likely a remnant of a past Cloud Shell or Firebase trial—and forces the Enterprise authentication flow.
- It auto-populates
geminicodeassist.projectentry in.vscode/settngs.jsonwith this shadow project ID. Refuses to allow to delete it. Each time it is manually deleted, it restarts the Extension fully and re-adds it. It is consistently always the same ID.
- If I manually try to remove the project in the
settings.json, it autopopulates. Video of the behavior: https://youtu.be/UUJ4oWZzInQ
- It doesn't matter the environment, it will do this across any OS, no matte how many time you manually delete extension caches, clear settings, sign in/out.
- Clicking on
Select Project...in the bottom right menu will just show an empty list (I am not Enterprise and I do NOT use Google Cloud for anything). - If I
Create a project..., it will create a Google Cloud empty project, and allow me to use it's "Free Tier" eligibility plan, but expect additional payment for using "Entreprise" AI. I already paid for Individual AI Pro, and expect to use my local projects, not cloud. - There is not option ANYWHERE in the extension for forcing the switch to my "Individual AI Pro" license. The moment I sign in, it auto-locks into Enterprise Mode, while my Google Account is a Personal Account.
- Support deadlock: I have already contacted Google One Support. They say to contact Google Cloud support. I am not a Google Cloud user, and I have nothing with them, but Google Cloud REQUIRES payment for any contact with them. "Billing Support" doesn't work either, I do not have a Billing Account, this is a Personal Account with no Google Cloud usage.
- I have tried: - Resetting extension
- "Enable API"
- Setting other project through UI
- Manually messing with files
- Deleting everything
- Switching to completely different machines, even brand new VM images.
- Going to Google Account Permissions and revoking "Google Cloud" (never was there), "Cloud Code" (never was there) and "Gemini Code Assist & CLI" entirely.
- Completely deleted ~/.config/Code/User/globalStorage/googlecloud.geminicodeassist
Repro step:
- Subscribe to Gemini Advanced / AI Pro.
- Download
Gemini Code Assist - Click
Sign in - The browser authentication flow requests cloud-platform scopes immediately.
Sign insuccessful through web using my personal Google Account that has a historical "Shadow Project" (e.g., from using Cloud Shell once in the past) but NO active Google Cloud Billing account or active projects.
- Extension goes to ready state, but already shows error message with the Magically Appearing "ghost project" (it's always same ID):
- If no filesystem access is required in the promp-t, it completes:
- If I query ANY promt that requires filesystem, the chat interface hangs on "Working..." for 10 minutes (timeout loop) because it is attempting Remote RAG on a project I cannot access, ignoring Local RAG entirely.
- Eventually, it crashes with no error message:
- It also defaults to "Max Usage" when there was no usage since subscription! Account has NOT used any Gemini Pro features at the point of this screenshot:
Attaching logfiles of behavior during screenshots/recording of this error: