Overview

Introduction

AgentKit by Inngest

A TypeScript library to create and orchestrate AI Agents.

AgentKit

Learn the basics of AgentKit in a few minutes.

Quick start

Installation

Run AgentKit locally with live traces and logs.

Local development

Create agents to accomplish specific tasks with tools inside a network.

Agents

Extending the functionality of Agents for structured output or performing tasks.

Tools

Combine one or more agents into a Network.

Networks

Shared memory, history, and key-value state for Agents and Networks.

State

Customize how calls are routed between Agents in a Network.

Routers

Leverage different provider's models across Agents.

Models

Deploy your AgentKit networks to production.

Deployment

Deterministic state routing

Provide your Agents with MCP Servers as tools

MCP as tools

Use multi-steps tools to create more complex Agents.

Multi-steps tools

Configure retries for your AgentKit network Agents and Tool calls.

Configuring Retries

Configure capacity based on users or organizations.

Configuring Multi-tenancy

Enable your Agents to wait for human input.

Human in the Loop

A comprehensive guide to building AI Agents with AgentKit

The three levels of AI apps

Leveraging AgentKit's Agent concept to power a RAG workflow.

1. Explaining a given code file

Code Assistant v1: Explaining a given code file

Use AgentKit Tools and Custom Router to add agentic capabilities.

2. Complex code analysis

Code Assistant v2: Complex code analysis

Build a custom Agent Router to autonomously solve bugs.

3. Autonomous Bug Solver

Code Assistant v3: Autonomous Bug Solver

Develop Coding Agents using E2B Sandboxes as tools

E2B - Sandboxes for AI Agents

Using AgentKit with E2B

Develop AI Agents that can browse the web

Browserbase - AI Browsers

Using AgentKit with Browserbase

Provide your Agents with hundred of prebuilt tools to interact with

Smithery - MCP Registry

Examples

createAgent

createTool

createNetwork

Leverage a Network's State across Routers and Agents.

createState

Controlling the flow of execution between agents in a Network.

Router

Network Router

OpenAI Model

Configure Anthropic as your model provider

Anthropic Model

Configure Google Gemini as your model provider

Gemini Model

Grok Model

Recent releases, new features, and fixes.

Introduction

Agent

Network

Models

Introduction

Overview

Agent APIs

Network APIs

Introduction

Agent

Network

Models

​Overview

Agent APIs

Network APIs

Overview