| layout | default |
|---|---|
| title | Gemini CLI with Vertex AI |
The repository includes project settings, maintainer instructions, and release commands for Gemini CLI. The project settings require Vertex AI authentication. Routine file edits, repository inspection, dependency setup, offline tests, and pre-commit checks are pre-approved. Commands that commit, tag, push, publish, or run live tests still require explicit approval.
- Node.js 20 or later.
- Gemini CLI installed with
npm install -g @google/gemini-cli. - Google Cloud CLI with Application Default Credentials (ADC).
- A Google Cloud project with billing and the Vertex AI API enabled.
- The Vertex AI User role (
roles/aiplatform.user) on that project.
Authenticate locally:
gcloud auth application-default login
gcloud services enable aiplatform.googleapis.com --project YOUR_PROJECT_IDCopy the checked-in environment template and set the project and region:
Copy-Item .gemini/.env.example .gemini/.env.gemini/.env is ignored by Git. Do not put API keys or service-account JSON
in the repository. ADC is preferred for local maintainer work.
Install the project dependencies, then start Gemini from the repository root:
poetry install --all-extras
geminiTrust the folder when Gemini CLI asks so it can load the checked-in project
settings and custom commands. Use /memory show to confirm GEMINI.md is
loaded and /commands list to confirm the release commands are available.
Prepare a candidate without publishing it:
/release:prepare 0.4.0
Review the diff, then run the independent release verification:
/release:verify
Both commands deliberately stop before live Substack tests, commits, tags, pushes, GitHub releases, PyPI publication, and announcements. Authorize those steps explicitly only after reviewing the candidate and confirming CI passed.