Add manual input for gemini's project_id variable

[MaliNamNam]

Feature Type

New functionality

Description

An option that asks the user for the project id so that it could be manually set.

Use Case

Some gemini accounts just wont have any tier and needs manual creation of google cloud console project with the api enabled.

Additional Context

No response

Pre-submission Checklist

• I have searched existing issues to ensure this is not a duplicate

[mirrobot-agent]

@MaliNamNam Thank you for submitting this issue. I am now beginning my analysis and will report back shortly.

[mirrobot-agent]

Initial Analysis Report

Summary: The issue requests adding manual input capability for the project_id variable during Gemini CLI OAuth credential setup, which would help users whose accounts require manual Google Cloud Console project creation.

Issue Validation: Confirmed - This is a valid enhancement request for the Gemini CLI OAuth provider.

Reproducibility Assessment: Not applicable (feature request, not a bug).

Root Cause Analysis:
The current implementation automatically discovers the project_id through the following flow in gemini_auth_base.py:_discover_project_id() (lines 128-629):

1. Checks in-memory cache
2. Checks GEMINI_CLI_PROJECT_ID environment variable override
3. Checks persisted project_id in credential file metadata
4. Calls loadCodeAssist API endpoint
5. Triggers onboardUser flow for new accounts
6. Falls back to GCP Resource Manager project listing

The problem occurs when:

• User has no existing Google Cloud project
• The automatic onboarding flow fails or is incompatible with their account type
• The account requires manual project creation in Google Cloud Console

Current Workaround: Users must manually set GEMINI_CLI_PROJECT_ID environment variable BEFORE running OAuth setup, which requires:

1. Creating a project in Google Cloud Console manually
2. Enabling the Cloud AI Companion API
3. Setting the env var in .env file
4. Then running OAuth setup

Suggested Labels: enhancement, gemini_cli, oauth

Proposed Next Steps:

1. Modify setup_credential() in google_oauth_base.py:1480 or setup_new_credential() in credential_tool.py:1720 to:

   • Prompt user for project_id after OAuth authentication
   • Make this prompt conditional (only if auto-discovery fails or user chooses manual mode)
   • Store the provided project_id in credential metadata before calling _post_auth_discovery()

2. Implementation approach (recommended):

   • Add an optional project_id parameter to CredentialSetupResult dataclass (already exists at line 58)
   • In credential_tool.py, after OAuth completes, check if discovery succeeded
   • If no project_id was discovered, prompt: "Enter your Google Cloud Project ID (or press Enter to skip auto-discovery):"
   • Pass the user-provided value to the credential file metadata

3. Files to modify:

   • src/rotator_library/credential_tool.py - Add interactive prompt in setup_new_credential()
   • src/rotator_library/providers/gemini_auth_base.py - Accept user-provided project_id in _discover_project_id() flow

Missing Information (if any):

• None - the issue description is clear about the use case

Investigation Warnings

• No warnings - all investigation commands completed successfully

This analysis was generated by an AI assistant.

[Mirrowel]

@mirrobot-agent make a PR to implement this suggestion. Base it on dev branch.
It needs to ask the user gcp project ID and provide instructions on how to manually enable code assist. That instruction will be in a new .md file in /docs. So it needs to tell the user to follow the instructions with a hyperlink to the doc in the repository docs folder.

Leave the doc itself placeholder so i can manually add the instructions.

This new step should only trigger if project ID is still not discovered after onboarding. It needs to auto-detect the tier after it is added using that project, and provide selection of the tier manually from the list of known tiers if tier can't be detected automatically.

This is entirely new functionality, so pay attention.

After you submit the PR - make sure PR body is populated. You made mistakes earlier with PR body messages that were missing.
Make sure to document the changes and what is done for the review, why, link issues, etc.

[mirrobot-agent]

@Mirrowel, I'm starting implementation of the manual project_id input feature. I'll create a PR based on the dev branch that:

1. Adds manual project_id input prompt after OAuth if auto-discovery fails
2. Creates a placeholder docs file with instructions for enabling Code Assist
3. Adds manual tier selection if auto-detection fails
4. Links the documentation in the user prompt

I'll report back with the PR link once complete.

[mirrobot-agent]

@Mirrowel, I have successfully implemented and committed the requested changes.

Summary

Implemented manual project_id input functionality for Gemini CLI OAuth when auto-discovery fails after onboarding.

Key Changes Made

1. Manual Input Method (src/rotator_library/providers/gemini_auth_base.py)

• Added _prompt_for_manual_setup() that:
  • Checks for interactive mode (stdin.isatty())
  • Displays instructions with link to docs
  • Prompts for project_id and tier selection
  • Returns dict with user input

2. Integration in Discovery Flow

• Modified _discover_project_id() to call manual prompt after:
  • Cache checks
  • Environment variable checks
  • loadCodeAssist API calls
  • onboarding flow
  • GCP project listing fallback
• Only triggers in interactive mode
• Falls back to original error if manual input unavailable

3. Documentation

• Created docs/gemini-cli-manual-code-assist-setup.md
• Placeholder for manual Code Assist setup instructions
• Structure ready for you to add step-by-step guide

4. Tier Selection

• KNOWN_GEMINI_TIERS constant with: free-tier, legacy-tier, standard, premium, enterprise
• User can select from list or choose "Other (unknown)"
• Auto-detection attempted but falls back to manual selection

The Fix

The implementation follows the exact flow you requested:

1. Auto-detect first: Tries all automatic methods (including onboarding)
2. Manual prompt on failure: Only after onboarding fails
3. Documentation link: Shows docs/gemini-cli-manual-code-assist-setup.md
4. Tier selection: Provides list of known tiers for manual selection
5. Persistence: Saves project_id and tier to credential metadata

Pull Request Created

The changes are now ready for review in:
#152

PR Details:

• Branch: feature/gemini-manual-project-input
• Base: dev
• Files changed: 2 (gemini_auth_base.py, docs/gemini-cli-manual-code-assist-setup.md)
• Additions: 220 lines
• Deletions: 1 line

Next Steps

1. Review PR feat(gemini): Add manual project_id input when auto-discovery fails #152 for code quality and logic
2. Complete documentation in docs/gemini-cli-manual-code-assist-setup.md with:
   • Step-by-step Google Cloud Console instructions
   • Screenshots
   • API enablement steps
   • Troubleshooting guide
3. Test with an account that has no existing projects

Warnings

• None - all changes completed successfully
• LSP errors shown are pre-existing import resolution issues in the dev branch

This update was generated by an AI assistant.