Google AI Studio deployment stuck with revoked GEMINI_API_KEY reference — Need to update secret without losing custom domain

Gemini API
,
1

Hello Google AI Studio team,

I’m experiencing a deployment configuration issue with my application, RMW Trading Journal, and would appreciate technical guidance.

Project details:

  • Application: RMW Trading Journal

  • Applet ID: 853d6431-56e9-4cae-803d-a12e822963..

  • Cloud Run service: rmw-trading-journal

  • Region: europe-west2

  • Custom domain: rmw-trading-journal.ai.studio

  • Current deployment status: Failed

  • Last failed revision: rmw-trading-journal-00003-454

Problem:

The original Gemini API key, whose reference ends in ...wTEA, was revoked and deleted after being exposed.

I subsequently created a new Gemini API key and updated the GEMINI_API_KEY secret in Google AI Studio under Settings → Secrets.

However, the Publish dialog still displays the reference to the old key (...wTEA). There is no visible option to edit or replace this reference, even when clicking the key name or its associated field.

I have also confirmed that the old key no longer exists as an active key in my API Keys management workflow.

Deployment issue:

The previous Cloud Run revision also failed during startup with the following error:

SyntaxError: Named export 'Request' not found. The requested module 'express' is a CommonJS module.

The source code has since been corrected to use:

import express from 'express';
import type { Request, Response } from 'express';

The workspace build and local server startup were subsequently reported as successful, but the corrected source has not yet been successfully deployed.

Important limitation:

My account does not have the run.services.update permission on the managed Google Cloud project, so I cannot update the Cloud Run service configuration directly.

What I need help with:

  1. Can the existing deployment’s GEMINI_API_KEY secret reference be refreshed to use the newly configured key?

  2. Is there a supported way to update this secret binding through Google AI Studio without unpublishing the application?

  3. Can the corrected source code be redeployed while preserving the existing application and custom domain?

  4. If this requires an internal intervention, could the relevant engineering team assist?

Critical requirement:

I must preserve the existing custom domain:

rmw-trading-journal.ai.studio

I do not want to use Unpublish or recreate the application if doing so would release the domain or cause data loss.

I would appreciate clear instructions on the safest supported recovery procedure.

Thank you for your help.

2

Hi @Eduardo_Lima, The Publish panel displays the API key from your last deployment until a new revision is published. Do not click Unpublish—instead, confirm an active API key exists on your selected project on the API Keys page, then click Republish in the Publish panel to deploy your updated code and refresh the key binding in place while keeping your custom domain. If Republish fails or still shows the old key afterward, please share your Project ID and a screenshot of the error via Google AI Developer Forum: Share Details About Your Issue.