google-cloud-auth
Provides expert guidance on authenticating and authorizing to Google Cloud services and APIs, covering human users, service identities, Application Default Credentials (ADC), and best practices for secure access.
- License 许可
- MIT
- 最近更新
- 2026-11-14
- Stars
- ★ 3.3w
- Forks
- 3.9k
01 前言 Front Matter
- name
- google-cloud-auth
- metadata
- { "source": "google/skills (Apache 2.0)" }
- description
- Provides expert guidance on authenticating and authorizing to Google Cloud services and APIs, covering human users, service identities, Application Default Credentials (ADC), and best practices for secure access.
03 SKILL.md 正文
Authenticating to Google Cloud
Authentication is the process of proving who you are. In Google Cloud, you represent a Principal (an identity like a user or a service). This is the first step before Authorization (determining what you can do).
Authentication
Clarifying Questions for the Agent
Before providing a specific solution, clarify the following with the user:
- Who or what is authenticating? (A human developer, a local script, or an application running in production?)
- Where is the code running? (Local laptop, Compute Engine, GKE, Cloud Run, or another cloud like AWS/Azure?)
- What is the target? (A Google Cloud API like Storage/BigQuery, or a custom application you built?)
- Are you using a high-level client library? (e.g., Python, Go, Node.js libraries usually handle ADC automatically.)
Human Authentication
For users to access Google Cloud, they need an identity that Google Cloud can recognize.
Types of User Identities
Google Cloud supports several ways to configure identities for your internal workforce (developers, administrators, employees):
- Google-Managed Accounts: You can use Cloud Identity or Google Workspace to create managed user accounts. These are called managed accounts because your organization controls their lifecycle and configuration.
- Federation using Cloud Identity or Google Workspace: You can federate identities to allow users to use their existing identity and credentials to sign in to Google services. Users authenticate against an external identity provider (IdP), but you must keep accounts synchronized into Google Cloud using tools like Google Cloud Directory Sync (GCDS) or an external authoritative source like Active Directory or Microsoft Entra ID.
- Workforce Identity Federation: This lets you use an external IdP to authenticate and authorize a workforce using IAM directly. Unlike standard federation, you do not need to synchronize user identities from your existing IdP to Google Cloud identities. It supports syncless, attribute-based single sign-on.
Methods of Access for Developers and Administrators
Used for interacting with Google Cloud resources and APIs during development and management.
- Google Cloud Console: The primary web interface. You authenticate using your Google Account (Gmail or Google Workspace).
- gcloud CLI (
gcloud auth login): Used to authenticate the CLI itself so you can run management commands (e.g.,gcloud compute instances list). It uses a Credential (like an OAuth 2.0 refresh token) stored locally. - Local Development with App Default Credentials
(ADC)
(
gcloud auth application-default login): This is different from CLI auth. It creates a local JSON file that Google Cloud Client Libraries (Python, Java, etc.) use to act as "you" when you run code on your laptop. - Service Account
Impersonation:
For security reasons, developers should avoid downloading Service Account
keys entirely. Instead, they should authenticate as humans (
gcloud auth login) and use Service Account Impersonation to run CLI commands or generate short-lived credentials. This is a critical best practice for local development and troubleshooting.
For End-Users and Customers
Used when a human (who is not a developer) needs to access a web application you've deployed on Google Cloud. Note: These are distinct from workforce identities.
- Identity-Aware Proxy (IAP): Acts as a central authorization layer for web applications. It intercepts web requests and verifies the user's identity (via Google Workspace, Cloud Identity, or external providers) before letting them reach the application. It's often used to protect internal apps without a VPN, or secure customer portals.
- Identity Platform: A Customer Identity and Access Management (CIAM) solution for adding consumer sign-in (email/password, phone, social) directly into the code of your custom-built applications.
Service-to-Service Authentication
When code runs in production, it should use a Service Account rather than a human user account.
Service Accounts and Service Agents
- Service Account: A special identity intended for non-human users. It's like a "robot identity" with its own email address.
- Service Agent: A service account managed by Google that allows a service (like Pub/Sub) to access your resources on your behalf.
Best Practice: Attaching Service Accounts
Instead of using Service Account Keys (dangerous JSON files), you should attach a custom service account to the Google Cloud resource. The resource's environment then provides a Token (a short-lived digital object) via a local metadata server.
- Compute Engine: Assign a service account during VM creation.
- Cloud Run: Assign a service account in the service configuration.
Special Cases & Advanced Topics
Kubernetes Engine (GKE)
Use Workload Identity Federation for GKE to map Kubernetes identities to IAM principal identifiers. This grants specific Kubernetes workloads access to specific Google Cloud APIs. Learn more here.
External Workloads (Workload Identity Federation)
For code running outside Google Cloud (e.g., AWS, Azure, or on-prem), do not use keys. Instead, use Workload Identity Federation to exchange an external token (like an AWS IAM role) for a short-lived Google Cloud access token.
API Keys
API keys are encrypted strings used for public data (e.g., Google Maps) or simplified access like Vertex AI Express Mode, which allows fast testing of Gemini models without complex setup. Both humans and services (e.g., Cloud Run-based AI agent) can use API keys, for the services that support it.
Note: API keys should be restricted to specific APIs and projects to minimize security risks. Store API keys in a secrets manager like Secret Manager to prevent accidental exposure.
OAuth 2.0 Access Scopes
While IAM is the modern way to handle authorization, legacy Compute Engine VMs and GKE node pools still rely on Access Scopes alongside IAM. If a VM's scope is restricted, the attached service account will fail to make API calls even if it has the correct IAM permissions. Check this first if attached service accounts are failing unexpectedly.
Short-Lived Credentials
The underlying mechanism for impersonation and secure service-to-service communication is the IAM Service Account Credentials API. This API generates short-lived access tokens, OpenID Connect (OIDC) ID tokens, or self-signed JSON Web Tokens (JWTs) dynamically, removing the need for static credentials.
Authorization
After Authentication, Google Cloud uses Identity and Access Management (IAM) to determine what the authenticated principal can do.
- Allow Policy: A record that binds a Principal to a Role on a Resource.
- Predefined
Roles:
Prebuilt roles like
roles/storage.objectViewerorroles/bigquery.dataEditor. Always try to use these first. - Custom Roles: User-defined collections of specific permissions if predefined roles are too broad.
Examples
Human-to-Service (Local Python Development)
- Authn: Run
gcloud auth application-default loginto create local credentials (ADC). - Authz: Grant your email the
roles/storage.objectViewerrole on a bucket. - Code: Use the Python
storage.Client(). It automatically finds your local credentials via ADC. Note: ADC searches in a specific order—first checking theGOOGLE_APPLICATION_CREDENTIALSenvironment variable, then the local gcloud JSON file, and finally the attached service account metadata server.
Service-to-Service (Cloud Run to Cloud SQL)
- Authn: Attach a custom Service Account to your Cloud Run service.
- Authz: Grant that Service Account the
roles/cloudsql.clientrole on the project. - Code: The Cloud Run environment provides the token automatically to the connection driver.
Calling a Custom Application (OIDC)
When calling a private Cloud Run service from another service, the caller
generates a Google-signed OpenID Connect (OIDC) ID Token and passes it in
the Authorization: Bearer <TOKEN> header.
Validation Checklist
- [ ] Is the user running code locally? Suggest
gcloud auth application-default loginor Service Account Impersonation. - [ ] Is the user attempting to use Service Account keys locally? Strongly discourage this and recommend impersonation.
- [ ] Is the user running in production? Recommend attaching a custom, least-privilege service account, NOT using keys.
- [ ] Is the user relying on the Compute Engine Default Service Account? Recommend creating a custom service account instead.
- [ ] Is the user running on another cloud? Recommend Workload Identity Federation.
- [ ] Is the user calling a custom app? Recommend OIDC ID Tokens.
- [ ] Has the user restricted their API Keys? Check for appropriate API Key Restrictions.
References
04 同仓库更多 Skills
wealthville-pool-scoring
Score DeFi liquidity pools before providing liquidity. Returns a 0-100 pool score and an Enter/Hold/Exit/Reduce/Avoid verdict with calibrated confidence for ~68,800 Solana pools and 575 EVM pools, plus a public miss-inclusive track record. Use when asked whether a pool is worth LPing into, which pools look best right now, or how accurate the scores have actually been.
technical-debt-auditor
Scans a codebase and produces a quantified technical debt report with a prioritized remediation plan. Analyzes dead code, TODO/FIXME/HACK markers, dependency staleness, complexity hotspots, code duplication signals, test coverage gaps, unused exports, and outdated patterns. Use this skill when the user asks to audit their codebase for tech debt, wants a health check of their code quality, says things like "where should I focus refactoring", "find code smells", "audit my codebase", "what technical debt do we have", "prioritize cleanup work", or asks about code quality metrics. Also trigger when the user mentions "code rot", "legacy cleanup", "refactoring priorities", or wants to understand the overall health of a project before a major effort. Trigger even if the user doesn't say "technical debt" explicitly — any request to systematically assess code quality and identify improvement priorities belongs here.
readonly-sql-guard
Make an agent's database access provably read-only. Use when an agent runs caller-supplied SQL against a production/ERP/analytics database (a "query" MCP tool, text-to-SQL, an admin console), or when reviewing DB-access code for write/DDL/side-effect/transaction-escape risks. Checks what a statement CALLS, not just its shape.
web-artifacts-builder
Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.
webapp-testing
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
discernment-nudge
After you give a substantive answer or draft that the user may act on — advice or recommendations, drafted artifacts such as goals, plans, pitches, proposals, or emails, estimates or projections, analysis or interpretation of data, factual claims they may rely on, or a multi-step argument — invoke this skill BEFORE finalizing your reply and then, if it applies, append 2-3 short follow-up questions, each tied to something specific in what you just produced, that help the user check key facts, probe the reasoning or assumptions, and notice missing context. Do this at most once per conversation. Skip it when the user asked a trivial how-to or simple lookup, wants a purely educational explanation, asked you only to format, convert, or assemble a file from content they provided, is writing code they will run, is doing creative writing or casual chat, or already asked you to double-check, cite, or review — the skill file explains these boundaries and the exact output format.