AgentFlowAcademy

Track · For developers

Agent Development Kit

Everything in the official ADK docs, animated: agents, workflows, tools, context, runtime and deployment.

Pro 60 lessons · 11 modules · 60 live now

Module 1 Get started

  1. 01 ADK technical overview Agents, Runner, Sessions, Events: the big picture. Start →
  2. 02 Install and project layout Python, TypeScript, Go, Java and Kotlin. Start →
  3. 03 Run your agent adk run, adk web, the API server and agents-cli. Start →
  4. 04 Agent Config and Visual Builder Agents in YAML, and building them visually. Start →
  5. 05 What's new in ADK 2.0 The graph-based workflow runtime, and migrating from 1.x. Start →

Module 2 Agents and models

  1. 06 LLM agents Instruction, description, model, output schema and output_key. Start →
  2. 07 Gemini models in ADK Choosing and configuring Gemini. Start →
  3. 08 Other models Claude, Gemma, LiteLLM, Ollama, vLLM, OpenAI, LiteRT-LM. Start →
  4. 09 Model routing and AI gateways Routing between models; Apigee AI Gateway. Start →
  5. 10 Managed agents and service registration Managed agents, and plugging in your own session and memory backends. Start →

Module 3 Workflows and multi-agent systems

  1. 11 Sequential, Parallel and Loop The three template workflows, side by side. Start →
  2. 12 Graph-based workflows Nodes, routes and deterministic control. Start →
  3. 13 Data handling between nodes How outputs flow from one node to the next. Start →
  4. 14 Dynamic workflows Loops and branches decided in code. Start →
  5. 15 Human input in workflows Pausing for a person, then resuming. Start →
  6. 16 Retries and timeouts RetryConfig and failing safely. Start →
  7. 17 Collaborative workflows Coordinators, sub-agents and agent routing. Start →
  8. 18 Sub-agent delegation vs agents as tools Handing off vs calling for help. Start →
  9. 19 Workflow patterns Proven shapes for real multi-agent systems. Start →
  10. 20 Deferred scheduling Running work later. Start →

Module 4 Tools

  1. 21 Function tools Plain functions, docstrings and tool schemas. Start →
  2. 22 Tool performance and limits What slows tools down, and the hard limits. Start →
  3. 23 Tool authentication OAuth, API keys and service accounts. Start →
  4. 24 Action confirmations Asking a human before a risky call. Start →
  5. 25 OpenAPI tools A whole REST API as tools, from one spec. Start →
  6. 26 MCP tools Model Context Protocol servers as toolsets. Start →
  7. 27 Your agent as an MCP server Exposing ADK agents to other MCP clients. Start →
  8. 28 Gemini built-in tools Google Search, code execution and computer use. Start →
  9. 29 Google Cloud tools BigQuery, Spanner, Bigtable, Pub/Sub, Application Integration, API Hub. Start →
  10. 30 Skills for ADK agents Packaged know-how an agent can load. Start →
  11. 31 The integrations ecosystem Third-party MCP tools, plugins and runners. Start →

Module 5 Context, sessions and memory

  1. 32 Agent context What each callback and tool can see. Start →
  2. 33 Sessions One conversation, its events and its lifecycle. Start →
  3. 34 State Session, user, app and temp prefixes. Start →
  4. 35 Memory Remembering across sessions. Start →
  5. 36 Artifacts Files and binary data, versioned. Start →
  6. 37 Events The event stream that drives everything. Start →
  7. 38 Context caching and compression Cutting cost and fitting long histories. Start →
  8. 39 Rewind and migrate sessions Going back in time; moving storage. Start →

Module 6 Control and safety

  1. 40 Callbacks Before and after the agent, the model and each tool. Start →
  2. 41 Callback patterns Guardrails, caching and logging in practice. Start →
  3. 42 Plugins Cross-cutting behaviour for a whole app. Start →
  4. 43 Safety and security Guardrails, Model Armor and least privilege. Start →

Module 7 Runtime

  1. 44 The event loop How the Runner and agents hand control back and forth. Start →
  2. 45 Runtime config Tuning a run. Start →
  3. 46 Cancel and resume runs Stopping safely and picking up later. Start →
  4. 47 Ambient agents Agents that run in the background. Start →

Module 8 Live and voice agents

  1. 48 Choosing an architecture Cascade or speech-to-speech. Start →
  2. 49 Audio, video and live sessions Streaming in both directions. Start →
  3. 50 Tools and guardrails in live agents Doing things while someone is talking. Start →

Module 9 Grounding

  1. 51 Google Search grounding Answers with citations from the web. Start →
  2. 52 Grounding with your data Agent Search and file retrieval. Start →

Module 10 Agent-to-agent and UIs

  1. 53 Introduction to A2A How agents talk across services. Start →
  2. 54 Exposing and consuming A2A agents Agent cards, servers and clients. Start →
  3. 55 Agent UIs: A2UI and AG-UI Agents that render rich interfaces. Start →

Module 11 Deploy and observe

  1. 56 Deploy to Agent Runtime Google Cloud's managed home for ADK agents. Start →
  2. 57 Deploy to Cloud Run and GKE Containers you control. Start →
  3. 58 OpenTelemetry in ADK The traces and spans ADK emits, and where to send them. Start →
  4. 59 Logging and metrics Seeing inside a production agent. Start →
  5. 60 Third-party observability Phoenix, Langfuse, Arize, Datadog and others. Start →