| Create a basic LLM agent | reference/getting-started/python-quickstart.md or typescript-quickstart.md / go-quickstart.md / java-quickstart.md |
| Understand LlmAgent parameters (name, model, tools, instructions, output_key) | reference/agents/llm-agents.md + language sidecar |
| Build a sequential pipeline of agents | reference/agents/workflow/sequential-agents.md |
| Build a loop / retry pattern | reference/agents/workflow/loop-agents.md |
| Run agents in parallel | reference/agents/workflow/parallel-agents.md |
| Build a multi-agent system with routing | reference/agents/multi-agent-systems.md + language sidecar |
| Create a custom agent (subclass BaseAgent) | reference/agents/custom-agents.md + language sidecar |
| Write a function tool with ADK patterns | reference/tools/function-tools-overview.md + language sidecar |
| Use LongRunningFunctionTool or AgentTool | reference/tools/function-tools-overview.md + language sidecar |
| Connect an MCP server as tools | reference/tools/mcp-tools.md + reference/tools/mcp-tools-python.md |
| Expose an OpenAPI spec as tools | reference/tools/openapi-tools.md |
| Add OAuth/API key auth to tools | reference/tools/tool-authentication.md + reference/tools/tool-authentication-python.md |
| Add human-in-the-loop confirmations | reference/tools/action-confirmations.md |
| See all built-in & third-party integrations | reference/tools/tools-and-integrations.md |
| Read/write session state, use state prefixes | reference/sessions/state.md + language sidecar |
| Use output_key to pass data between agents | reference/sessions/state.md + language sidecar |
| Use persistent sessions (DatabaseSessionService) | reference/sessions/sessions.md |
| Add long-term memory (MemoryService) | reference/sessions/memory.md + language sidecar |
| Add callbacks (before/after model/tool/agent) | reference/callbacks/types-of-callbacks.md + language sidecar |
| Callback patterns and best practices | reference/callbacks/callback-patterns.md |
| Configure models (Gemini, Claude, Vertex, Ollama, LiteLLM) | reference/models/models-overview.md |
| Use Claude (Anthropic) as the model | reference/models/claude.md |
| Use Vertex AI hosted models | reference/models/vertex-ai.md |
| Use Ollama / local models | reference/models/ollama.md |
| Build a streaming / realtime agent | reference/streaming/streaming-overview.md + reference/streaming/part1-intro-to-streaming.md + reference/streaming/part1-intro-to-streaming-python.md |
| Stream audio, images, or video | reference/streaming/part5-audio-images-video.md |
| Use Google Search grounding | reference/grounding/google-search-grounding.md |
| Use Vertex AI Search grounding | reference/grounding/vertex-ai-search-grounding.md |
| Work with artifacts (files, blobs) | reference/components/artifacts.md |
| Run the agent (CLI / web UI / API server) | reference/runtime/runtime-overview.md |
| Deploy to Cloud Run | reference/deployment/cloud-run.md |
| Deploy to Vertex AI Agent Engine | reference/deployment/agent-engine.md |
| Evaluate agent quality / write eval tests | reference/evaluation/evaluation-overview.md |
| Expose an agent via A2A protocol | reference/a2a/quickstart-exposing-python.md |
| Consume a remote agent via A2A | reference/a2a/quickstart-consuming-python.md |
| Enable tracing / logging / observability | reference/observability/observability-overview.md |
| Use context caching to reduce cost | reference/context/context-caching.md |
| Compress long context windows | reference/context/context-compression.md |
| CLI commands reference | reference/reference/cli-reference.md |
