Agent skill · software engineering · jeremylongshore
google-cloud-agent-sdk-master
Build, evaluate, deploy, publish, or modernize Google ADK agent applications with Google's maintained agents-cli workflow. Use when a request mentions Google Agent Development Kit, agents-cli, Gemini Enterprise Agent Platform, Agent Runtime, Cloud Run or GKE agent deployment, or migration from Agent Starter Pack. For Claude Code agents/*.md definitions, use agent-creator instead. Trigger with "build a Google ADK agent", "deploy this ADK app", or "migrate Agent Starter Pack".
Why this skill is useful
Provides specific Google Cloud Agent Development Kit patterns and validation steps that the AI wouldn't reliably generate on its own.
What it needs
About 3k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Google Cloud ADK Application Workflow Overview Guide a Google ADK application through its current lifecycle: inspect, scaffold, run, evaluate, deploy, publish, and observe. Prefer Google's actively maintained agents-cli. Agent Starter Pack is in maintenance mode and should be treated as a migration source, not the default for new work. This skill creates Google-hosted agent applications. It does not create Claude Code subagent definition files; route those requests to agent-creator. Prerequisites Identify whether the target is new, an existing ADK app, or an Agent Starter Pack project that needs migration. Confirm the local project path and read its instructions. For current-version questions, use WebFetch on official Google documentation or WebSearch restricted to official Google domains. For existing code, use Glob, Grep, and Read to establish its actual framework, configuration, tests, and deployment lane before recommending work. Do not require Google Cloud for local development; AI Studio credentials may support a local-only workflow. Cloud deployment does require a Google project. Instructions 1. Classify the request as build, enhance, evaluate, deploy, publish, observe, or migrate. State which lifecycle stage is in scope. 2. Inspect the local repository before proposing changes. Preserve working agent code, evaluation datasets, Terraform, and CI/CD during migrations. 3. Verify current CLI syntax from the installed command or official docs. Start from these maintained entry points: 4. For implementation guidance, define the ADK agent, tools, state, callbacks, error contracts, and tests before selecting infrastructure. Use the detailed lifecycle reference for current command families. 5. Validate locally with the generated project's own commands. Current agents-cli releases expose install, lint, run, and the eval command group. Do not claim success without the actual command receipt. 6. For deployment, distinguish Agent Runtime, Cloud Run, and GKE. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/google-cloud-agent-sdk-master--8871d3