google/adk-web

Agent Development Kit Web (adk web) is the built-in developer UI that is integrated with Agent Development Kit for easier agent development and debugging.

Agent Development Kit Web UI (ADK WEB)

What it is – A browser‑based developer console that ships with Google’s Agent Development Kit (ADK). ADK is a modular framework for building, deploying, and orchestrating AI agents (think chat‑bots, autonomous assistants, or workflow‑orchestrating agents). The Web UI gives you a visual front‑end for creating agents, inspecting events, tracing execution, viewing artifacts and evaluation results, and debugging the underlying ADK runtime.

Key pieces

  • Frontend – An Angular single‑page app (served with npm run serve). It talks to the ADK backend via a proxy, so the UI can be run locally at http://localhost:4200.
  • Backend – The ADK runtime itself, which you run with the adk web command (installed from the google-adk Python or Java packages). The UI forwards API calls to this process.
  • Integration – Works out‑of‑the‑box with the Gemini model stack but is model‑agnostic; you can point it at any ADK‑compatible agent implementation.

How to get started

  1. Install Node, npm, and the Angular CLI.
  2. Install the ADK Python or Java SDK (google-adk).
  3. Clone this repo, run npm install, then npm run serve.
  4. In another terminal start the backend with adk web.
  5. Open http://localhost:4200 and you’ll see the UI with tabs for Events, Tracing, Artifacts, Evaluations, and an Agent Builder.

What you can do in the UI

  • Events – Browse a timeline of agent‑generated events.
  • Tracing – Visualise call‑graphs and execution flow of an agent run.
  • Artifacts – Inspect files, logs, or other outputs produced by the agent.
  • Evaluations – View metrics and test results for your agent.
  • Agent Builder & Assistant – Drag‑and‑drop style interface for assembling agent components and interacting with a live assistant.

Who it’s for – Developers building AI agents who want a richer debugging experience than raw logs. Because it’s part of the ADK ecosystem, it fits naturally into Google‑centric workflows but can be used with any ADK‑compatible model.

Contribution – Follow Google’s ADK contribution guide; tests need a special initTestBed call for Google’s internal CI.

License – Apache 2.0.

Happy Agent Building!

Related

  • Project
    Lamatic/AgentKit

    AgentKit is an open‑source SDK that provides a catalog of 70 ready‑made “kits” (full apps, pipelines, or templates) for building, testing and deploying AI agents. Kits are defined by a simple config file, can be edited in a visual flow studio, and are shipped as Next.js apps that run server‑less. The collection covers support triage, code review, RAG chat, hiring assistants, legal/medical bots, and many other business‑focused agents, making it a turnkey platform for creating reliable, enterprise‑grade AI agents.

  • Project
    google/adk-go

    Google's Agent Development Kit for Go is an open-source, code-first framework for building, evaluating, and deploying modular AI agents. It applies software engineering practices—such as versioning and testability—to agent development, and is model-agnostic while optimized for Gemini.

  • Project
    googleapis/python-aiplatform

    The Python SDK for the Gemini Enterprise Agent Platform (formerly Vertex AI) allows developers to build, deploy, and optimize AI agents, manage prompts, and evaluate generative AI models.

  • Project
    OpenHands/software-agent-sdk

    A developer SDK providing APIs to build AI agents that can write code, manage files, and execute terminal commands in local or ephemeral workspaces.

  • Project
    agnt-gg/agnt

    AGNT is a local‑first, self‑hostable desktop app that bundles AI agents, visual workflows, long‑running goals, memory, plugins and a local HTTP API into one “agent operating system”. Built with Electron, Vue, Express and SQLite, it supports 15+ LLM providers, 60+ workflow nodes, plugin templates and SkillForge for self‑improving skills. Ideal for developers or teams that need durable, inspectable AI automation on‑premise.