Skip to content

Agent Runtime

Supported in ADKPython v0.1.0TypeScript v0.2.0Go v0.1.0Java v0.1.0

ADK provides several ways to run and test your agents during development. Choose the method that best fits your development workflow.

Ways to run agents

  • Dev UI

    Use adk web to launch a browser-based interface for interacting with your agents.

    Use the Web Interface

  • Command Line

    Use adk run to interact with your agents directly in the terminal.

    Use the Command Line

  • API Server

    Use adk api_server to expose your agents through a RESTful API.

    Use the API Server

  • Ambient Agents

    Build autonomous agents that process events, monitor systems, and respond asynchronously without human intervention.

    Use Ambient Agents

Technical reference

For more in-depth information on runtime configuration and behavior, see these pages:

  • Event Loop: Understand the core event loop that powers ADK, including the yield/pause/resume cycle.
  • Resume Agents: Learn how to resume agent execution from a previous state.
  • Cancel Agent Runs: Gracefully cancel running agent invocations using AbortSignal (TypeScript).
  • Runtime Config: Configure runtime behavior with RunConfig.