scaffold-google-adk

Scaffold Google ADK Service

Project name: $ARGUMENTS (default to "my-adk-service" if not provided)

Delegate to the google-adk skill for all patterns, templates, and reference files.

Pre-Scaffold Questions

Before writing any code or running any command, ask ALL of the following. Present DESIGN_SPEC.md for approval before running any scaffold command.

Always ask:

1. What problem will the agent solve? — Core purpose and capabilities
2. External APIs or data sources needed? — Tools, integrations, auth requirements
3. Safety constraints? — What the agent must NOT do, guardrails
4. Deployment preference? — Prototype first (recommended) or full deployment? If deploying: Agent Engine or Cloud Run?

Ask based on context:

• If retrieval or search over data mentioned (RAG, semantic search, vector search, embeddings, similarity search, data ingestion) → Datastore? Use --agent agentic_rag --datastore <choice>:
  • vertex_ai_vector_search — for embeddings, similarity search, vector search
  • vertex_ai_search — for document search, search engine
• If agent should be available to other agents → A2A protocol? Use --agent adk_a2a (see Option C below)
• If full deployment chosen → CI/CD runner? Choose github_actions (WIF-based, no PAT) or google_cloud_build (native GCP)
• If Cloud Run chosen → Session storage? cloud_sql for production (VertexAiSessionService); in_memory for dev/prototype
• If background job chosen (schedule/event trigger, no HTTP) → Use Option D (BackgroundJob pattern) — no FastAPI, no streaming

---

Steps

1. Read the google-adk skill (SKILL.md and reference files) before generating any code

2. Ask the user which setup approach they want:

   "Do you want: A) agent-starter-pack (production-grade — includes Terraform, Dockerfile, CI/CD, eval harness; recommended for real projects) B) Manual setup (simpler, no deployment scaffold — use for learning or quick prototypes)"