Google ADK Project Scaffolder

Scaffolds a new agent project from scratch with the Google Agent Development Kit (ADK) agents-cli tool, or enriches an existing project with deployment, CI/CD, and infrastructure layers. It clarifies what kind of agent the user wants (RAG, A2A, standard ADK), the session storage, and the deployment target; then maps those choices to the correct agents-cli scaffold create / enhance / upgrade flags.

When to use it: When you say "start a new ADK project", "create an agent project", "add deployment/CI-CD to my project", or "upgrade my project". Separate skills exist for writing the agent code itself (google-agents-cli-adk-code) and for live deployment operations (google-agents-cli-deploy).

What it produces: A structured Markdown setup report containing the command to run, a reasoned flag mapping for the chosen architecture, the generated project directory tree, the mandatory datastore provisioning sequence for RAG, and the next steps. It recommends a "prototype-first" approach: iterate quickly first with --prototype, then add deployment later with enhance.

Critical rules are embedded: do not mkdir the project directory in advance, clear session_type for the agent_runtime target, do not change the model, do not choose the CI/CD runner and Git repository without asking the user. The output is a Turkish report; the commands and code blocks are kept in their original form.