Google-ADK-Skills

ADK Agent-to-Agent (A2A) protocol expert covering remote agent communication, agent cards, HTTP-based delegation, and distributed agent systems. Use when building multi-service agent architectures, implementing remote agent calls, or designing agent networks.

ADK memory + embeddings across databases. Generate multimodal embeddings with gemini-embedding-2, persist agent sessions/memory to Cloud SQL, Postgres, Firestore, or SQLite, and store/retrieve vectors for RAG (pgvector, Firestore vector). Use when choosing a session/memory backend, wiring a database SessionService or MemoryService, generating embeddings, or building a vector store for retrieval on an ADK agent.

Migrate ADK agents from 1.x to 2.3. Covers breaking API changes (graph Workflow vs GraphAgent, McpToolset, BuiltInCodeExecutor, ctx.resume_inputs, agent modes), renamed imports, deploy CLI changes, model/thinking/embedding updates, and session/state incompatibility. Use when upgrading an ADK 1.x codebase to 2.3, auditing code for deprecated 1.x APIs, or resolving 2.3 import/attribute errors.

ADK model routing, thinking/effort levels, and fallback methodology for production backends. Covers the current Gemini 3.x model catalog (gemini-3.5-flash, gemini-3.1-flash-lite, gemini-3.1-pro-preview, gemini-3.1-flash-live-preview, gemini-embedding-2), thinking_level effort tiers (minimal/low/medium/high), deprecation of thinking_budget and sampling params, and primary→fallback chains. Use when selecting a model, tuning thinking/effort, routing by task complexity/cost, or designing model fallback for an ADK agent.

Central hub for building, testing, and iterating on ADK agents. Trigger this skill when the user wants to create a new agent, configure modes (task, single-turn), or build graph-based workflows.

DAG-based ADK workflow in Go using graph pattern. Use for Go backends needing multi-step agent pipelines.

Coordinator agent delegating to specialized sub-agents. Python. Use for multi-domain tasks requiring different expertise.

Programmatic ADK workflow using imperative control flow. Python. Use for complex logic where graph edges are insufficient.

DAG-based ADK 2.x Workflow template using START and edge tuples. Python. Use for multi-step pipelines with branching logic.

Iterative refinement using LoopAgent. Python. Generate → critique → improve → repeat until quality gate passes.

Fan-out parallelism using ParallelAgent. Python. Use for independent task batches that can run concurrently.

Linear pipeline using SequentialAgent. Python. Use for fixed-order processing chains.

ADK Agentic Production Workflow Builder. Design and generate production-grade multi-agent workflows (graph-based, dynamic, collaborative, template) with full scaffolding: agents, tools, infra, CI/CD, evals, security, and observability. Use when building ADK agent systems, composing multi-agent workflows, scaffolding agent projects, designing workflow architectures, or deploying ADK agents to production (Cloud Run, Agent Engine, GKE). Covers Python, Go, and TS backends with A2A and MCP protocol support.

ADK multi-agent systems expert covering agent composition, routing patterns, sequential/parallel execution, and supervisor architectures. Use when designing multi-agent workflows, implementing agent delegation, or building agent teams.

Build autonomous, self-reasoning ADK agents with proactive, goal-directed execution using the OODA loop (Observe, Orient, Decide, Act) and explicit planning. Use when a request asks for an agent that runs without per-step human follow-up, plans its own steps, or pursues a goal autonomously.

ADK backend architecture expert covering runtime event loops, session management, state handling, and server implementation patterns. Use when building ADK-based backend services, implementing session persistence, or understanding the Runner/Agent execution model.

ADK bidirectional streaming and Live API expert covering real-time audio/video, WebSocket communication, LiveRequestQueue, Gemini Live API, and native audio models (gemini-3.1-flash-live-preview). Use when building real-time voice agents, implementing streaming chat, or working with multimodal live interactions.

ADK configuration expert covering YAML agent definitions, environment variables, agent cards, and deployment configs. Use when setting up ADK projects, managing environment-specific settings, or defining agents declaratively.

Use when debugging ADK agents, inspecting sessions, testing agent behavior, troubleshooting tool calls, event flow issues, or diagnosing LLM/model problems.

ADK deployment specialist covering Cloud Run, GKE, Vertex AI Agent Engine, and Docker containerization. Use when deploying ADK agents to production, configuring cloud infrastructure, or setting up CI/CD pipelines for ADK applications.

Generate a production-ready expert ADK agent for any domain not covered by the standard personas. Define the domain expertise, interaction style, and knowledge sources, then emit a configured agent. Use when a request needs a bespoke specialist (e.g. a tax-law advisor, a marine-biology tutor) with no matching persona template.

Orchestrate stateful, multi-agent ADK workflows with LangGraph: graphs, state machines, conditional edges, and LLM-driven routing, using ADK tools as nodes. Use when a request mentions LangGraph, a state-machine/graph workflow, complex branching orchestration, or ADK↔LangGraph interop.

ADK LiteLLM integration expert covering 100+ LLM providers (OpenAI, Anthropic, Bedrock, OpenRouter), model switching, and cost optimization through provider flexibility. Use when integrating non-Google models, implementing model fallbacks, or optimizing costs across providers.

ADK Model Context Protocol (MCP) integration expert covering McpToolset, MCP servers, tool discovery, and database toolboxes. Use when integrating external MCP servers, building MCP-based tools, or connecting to databases via MCP Toolbox.

ADK memory and state management expert covering session state, long-term memory, persistence backends, and cross-session recall. Use when implementing conversation memory, persisting agent state, or building knowledge-retrieval systems.

Create specialized ADK agents instantly from 30+ pre-built persona templates, each with optimized instructions, recommended tools, and interaction style. Use when a request asks for a ready-made character/role agent (e.g. tutor, coach, analyst, support rep) rather than a bespoke build.

ADK prompt engineering expert covering agent instructions, few-shot examples, system prompts, and cost optimization strategies. Use when designing agent personalities, optimizing prompt tokens, or implementing few-shot learning patterns.

Add retrieval-augmented generation to ADK agents. Build knowledge-powered agents with either Pinecone (self-managed vector DB, embeddings, ingestion) or Vertex AI RAG (managed corpora). Use when a request mentions RAG, knowledge base, vector search, embeddings, document ingestion, or grounding an agent in a corpus.

Author new samples for the ADK Python repository. Use this skill when the user wants to create a new sample demonstrating a feature or agent pattern (e.g., dynamic nodes, standalone agents, fan-out/fan-in) or when adding examples to subdirectories under `contributing/`.

ADK tools expert covering custom function tools, Google built-in tools (Search, Code Execution), OpenAPI integrations, and tool composition. Use when creating custom tools, integrating external APIs, or configuring built-in Google tools.

ADK runtime architecture expert covering event loops, Runner execution model, callbacks, and the Reason-Act cycle. Use when understanding ADK internals, implementing custom callbacks, or debugging agent execution flow.

Set up a local development environment for the ADK Python project. Use when the user wants to get started developing, set up their environment, install dependencies, or prepare for contributing.

Coordinator + sub-agents sharing state. Go. Use for multi-agent collaboration with shared context.

Runtime agent composition with programmatic node creation. Go. Use for dynamic workflows that change based on input.

Fan-out to independent workers using ParallelAgent. Go. Use for concurrent search, multi-source fetch, parallel validation.

Fixed-order pipeline using SequentialAgent. Go. Use for deterministic Write→Review→Publish chains.

Production ADK agent server with AgentCard, A2A, auth middleware, and tools. TypeScript. Use as the agent-side template for TS deployments.

DAG-based routing with GraphAgent. TypeScript. Use for complex branching, conditional routing, fan-out/fan-in.

TypeScript client for consuming ADK workflows over HTTP/A2A. Use in Next.js, Express, or Cloud Functions frontends.

ADK architectural knowledge — graph orchestration, resumption, execution flow, node contracts, observability, and LLM context orchestration. Use this skill whenever you need to understand the architecture, event flow, or state management of the ADK system, or when designing or modifying core components. Triggers on "how does X work", "design of", "architecture of", "event flow", "resumption state", "checkpoint", "BaseNode", "NodeRunner".