agent-starter-pack GEMINI.md

Repository instructions for AI coding agents working on Google Cloud’s Agent Starter Pack, a template generator for creating agent projects. They explain its architecture, directory layout, and rules for making focused changes.

In plain words
What is it for?
Use them when modifying the Agent Starter Pack itself, especially when deciding where code belongs, following existing patterns, or preserving the project’s template structure.
Why use it?
They help an agent change the template generator safely without rewriting unrelated code or putting features in the wrong layer.

Instructions file for Gemini CLI

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/googlecloudplatform/agent-starter-pack/gemini-md
Clone the repo
git clone --depth 1 https://github.com/GoogleCloudPlatform/agent-starter-pack

Made for: Gemini CLI.

Per session 5,853 This file is loaded in full into every session.
When invoked 5,853 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.05853 $0.05853
Opus 5 $0.02926 $0.02926
Sonnet 5 $0.01171 $0.01171
Haiku 4.5 $0.00585 $0.00585

Measured yesterday against content hash e256e6fdc0fd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-starter-pack GEMINI.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

GEMINI.md · 723 lines

How it starts

The opening of the file, as written. The whole thing — 723 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Starter Pack - AI Coding Agent Guide

Scope: This document is for AI coding agents contributing to the Agent Starter Pack repository itself (the template generator). For guidance on working with generated projects, see llm.txt.

This document provides essential guidance, architectural insights, and best practices for AI coding agents tasked with modifying the Google Cloud Agent Starter Pack. Adhering to these principles is critical for making safe, consistent, and effective changes.


Core Principles for AI Agents

  1. Preserve and Isolate: Your primary objective is surgical precision. Modify only the code segments directly related to the user's request. Preserve all surrounding code, comments, and formatting. Do not rewrite entire files or functions to make a small change.
  2. Follow Conventions: This project relies heavily on established patterns. Before writing new code, analyze the surrounding files to understand and replicate existing conventions for naming, templating logic, and directory structure.
  3. Template-First Mindset: ASP is a template generator. The CLI should remain lean with good defaults. Most features belong in templates, not CLI code.
  4. Search Comprehensively: A single change often requires updates in multiple places. When modifying configuration, variables, or infrastructure, you must search across the entire repository, including:
    • agent_starter_pack/base_templates/ (core templates by language)
    • agent_starter_pack/deployment_targets/ (environment-specific overrides)
    • .github/ and .cloudbuild/ (CI/CD workflows)
    • docs/ (user-facing documentation)

Project Architecture Overview

4-Layer Template System

Template processing follows this hierarchy (later layers override earlier ones):

Layer Directory Purpose
1. Base agent_starter_pack/base_templates/<language>/ Core Jinja scaffolding (Python, Go, more coming)
2. Deployment agent_starter_pack/deployment_targets/ Environment overrides (cloud_run, gke, agent_engine)
3. Frontend agent_starter_pack/frontends/ UI-specific files
4. Agent agent_starter_pack/agents/*/ Agent-specific logic and configurations

Read the full file on GitHub · 723 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 723 lines · 5,853 tokens per session scan A e256e6fdc0fd

Subscribe to this mod's changes

agent-starter-pack GEMINI.md is an instructions file published in the GitHub repository GoogleCloudPlatform/agent-starter-pack (6,547 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 5,853 tokens to every session, about $0.0293 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.