Skip to main content
ADK-TS Logo

​
What is ADK-TS?

The Agent Development Kit (ADK) for TypeScript is a comprehensive framework for building sophisticated AI agents with multi-LLM support, advanced tool integration, memory systems, and flexible conversation flows. Built from the ground up for production use, ADK-TS enables developers to create intelligent, autonomous systems that can handle complex multi-step tasks.

​
Why ADK-TS?

ADK-TS simplifies the complexity of building AI agents by providing a unified interface across multiple LLM providers and a robust set of tools for creating production-ready systems.

Multi-provider support

Seamlessly integrate OpenAI, Anthropic, Google Gemini, and other leading providers with a single API

Extensible tools

Define custom tools with declarative schemas for intelligent LLM integration

Advanced reasoning

Complete reasoning loop implementation for complex task execution and planning

Memory systems

Persistent memory and context retention for learning from past interactions

Real-time streaming

Support for streaming responses and dynamic user interactions

Multi-agent orchestration

Sequential, parallel, and loop-based agent workflows for complex systems

​
Key features

ADK-TS provides a unified interface for working with multiple LLM providers:
  • OpenAI (GPT-4, GPT-3.5)
  • Anthropic (Claude)
  • Google (Gemini, Vertex AI)
  • AI SDK compatibility
  • Automatic provider registration
  • Fallback model support
Create custom tools with declarative schemas:
  • Zod schema validation
  • Type-safe tool definitions
  • Built-in tools for file operations, HTTP requests, and more
  • MCP (Model Context Protocol) integration
  • Function tools with automatic schema generation
Build different types of agents for various use cases:
  • LlmAgent: Single LLM-based agent for straightforward tasks
  • LoopAgent: Iterative planning and execution for complex workflows
  • ParallelAgent: Execute multiple agents concurrently
  • SequentialAgent: Execute agents in sequence
  • LangGraphAgent: Graph-based workflow orchestration
Persistent context and state management:
  • Pluggable memory services (in-memory, Vertex AI RAG)
  • Session management with database persistence
  • State management across conversations
  • Artifact storage and retrieval
Multiple execution environments for running code:
  • Local execution (development)
  • Container-based sandboxing
  • Vertex AI code execution
  • Built-in security controls

​
Quick example

Get started with a simple agent in just a few lines of code: 
import { AgentBuilder } from "@iqai/adk";

const response = await AgentBuilder.withModel("gemini-2.5-flash").ask(
  "What is the capital of France?"
);

console.log(response);

​
Get started

Ready to build your first AI agent? Choose your path:

Quickstart

Get a working agent running in minutes

Installation

Detailed installation guide and setup instructions

Core concepts

Learn the fundamental concepts of ADK-TS

API reference

Explore the complete API documentation

​
Community and support

Join our community to discuss ideas, ask questions, and share your projects:
ADK-TS is open source and licensed under the MIT License. We welcome contributions from the community!

Build docs developers (and LLMs) love

Get started for freeTalk to us