google-adk

Agent Development Kit

76 个版本 Python >=3.10
安装
pip install google-adk
poetry add google-adk
pipenv install google-adk
conda install google-adk
描述

Agent Development Kit (ADK) 2.0

License PyPI version Python versions PyPI downloads Unit Tests Docs

An open-source, code-first Python framework for building, evaluating, and deploying sophisticated AI agents with flexibility and control.

Important Links: Docs, Samples & ADK Web.


⚠️ BREAKING CHANGES FROM 1.x

This release includes breaking changes to the agent API, event model, and session schema. Sessions generated by ADK 2.0 are readable by ADK 1.28+ (extra fields will be ignored), but are incompatible with older 1.x versions.


🔥 What's New in 2.0

  • Workflow Runtime: A graph-based execution engine for composing deterministic execution flows for agentic apps, with support for routing, fan-out/fan-in, loops, retry, state management, dynamic nodes, human-in-the-loop, and nested workflows.

  • Task API: Structured agent-to-agent delegation with multi-turn task mode, single-turn controlled output, mixed delegation patterns, human-in-the-loop, and task agents as workflow nodes.

🚀 Installation

pip install google-adk

Requirements: Python 3.10+.

To install optional integrations, you can use the following command:

pip install "google-adk[extensions]"

The release cadence is roughly bi-weekly.

Quick Start

Beginner Note: ADK applications are built using two main classes: Agent (defines an AI's instructions, tools, and behavior) and Workflow (orchestrates agents and tasks in a graph-based flow).

Agent

from google.adk import Agent

root_agent = Agent(
    name="greeting_agent",
    model="gemini-2.5-flash",
    instruction="You are a helpful assistant. Greet the user warmly.",
)

Workflow

from google.adk import Agent, Workflow

generate_fruit_agent = Agent(
    name="generate_fruit_agent",
    instruction="Return the name of a random fruit. Return only the name.",
)

generate_benefit_agent = Agent(
    name="generate_benefit_agent",
    instruction="Tell me a health benefit about the specified fruit.",
)

root_agent = Workflow(
    name="root_agent",
    edges=[("START", generate_fruit_agent, generate_benefit_agent)],
)

Run Locally

# Interactive CLI
adk run path/to/my_agent

# Web UI (supports multi-agent directories or pointing directly to a single agent folder)
adk web path/to/agents_dir

📚 Documentation

🤝 Contributing

See CONTRIBUTING.md for details.

📄 License

This project is licensed under the Apache 2.0 License — see the LICENSE file for details.

版本列表
2.3.0 2026-06-18
2.2.0 2026-06-04
2.1.0 2026-05-23
2.0.0 2026-05-19
2.0.0b1 2026-04-21
2.0.0a3 2026-04-09
2.0.0a2 2026-03-27
2.0.0a1 2026-03-18
1.36.0 2026-06-22
1.35.2 2026-06-17
1.35.1 2026-06-15
1.35.0 2026-06-10
1.34.3 2026-06-02
1.34.2 2026-06-01
1.34.1 2026-05-22
1.34.0 2026-05-18
1.33.0 2026-05-08
1.32.0 2026-05-01
1.31.1 2026-04-21
1.31.0 2026-04-17
1.30.0 2026-04-13
1.29.0 2026-04-09
1.28.1 2026-04-02
1.28.0 2026-03-26
1.27.5 2026-03-26
1.27.4 2026-03-24
1.27.3 2026-03-23
1.27.2 2026-03-17
1.27.1 2026-03-13
1.27.0 2026-03-12
1.26.0 2026-02-26
1.25.1 2026-02-18
1.25.0 2026-02-11
1.24.1 2026-02-06
1.24.0 2026-02-05
1.23.0 2026-01-22
1.22.1 2026-01-12
1.22.0 2026-01-08
1.21.0 2025-12-12
1.20.0 2025-12-03
1.19.0 2025-11-20
1.18.0 2025-11-05
1.17.0 2025-10-22
1.16.0 2025-10-09
1.15.1 2025-09-26
1.15.0 2025-09-26
1.14.1 2025-09-15
1.14.0 2025-09-10
1.13.0 2025-08-27
1.12.0 2025-08-20
1.11.0 2025-08-14
1.10.0 2025-08-07
1.9.0 2025-07-31
1.8.0 2025-07-23
1.7.0 2025-07-16
1.6.1 2025-07-10
1.5.0 2025-06-25
1.4.2 2025-06-20
1.4.1 2025-06-19
1.3.0 2025-06-12
1.2.1 2025-06-04
1.2.0 2025-06-04
1.1.1 2025-05-29
1.1.0 2025-05-28
1.0.0 2025-05-20
0.5.0 2025-05-08
0.4.0 2025-05-02
0.3.0 2025-04-24
0.2.0 2025-04-18
0.1.1 2025-04-09
0.1.0 2025-04-09
0.0.5 2025-04-09
0.0.4 2025-04-09
0.0.3 2025-04-08
0.0.2 2025-04-08
0.0.1 2025-03-17