Handit.AI
  • Introduction
    • Introduction
    • Quick Start
  • Agent Creation
    • Overview
    • Interactive Agent Setup
    • Define Your Agent with a JSON Configuration
  • Code Integration
    • Overview
    • MCP Server Setup
    • Context Based Setup
    • Manual Setup
  • Best Practices
    • Input/Output Tracking
    • Service Initialization
    • Error Handling
Powered by GitBook
On this page
  • 1. Create Your Agent (2 minutes)
  • 2. Download and Set Up MCP Server (1 minute)
  • 3. Connect Your Code (1 minute)
  • What You Get
  • Need Help?
  1. Introduction

Quick Start

PreviousIntroductionNextOverview

Last updated 1 month ago

Get your Handit.AI agent up and running in just a few minutes. This guide walks you through creating your agent, connecting it to your code, and unlocking powerful real-time monitoring.

1. Create Your Agent (2 minutes)

Start by designing your AI agent’s logic using Handit.AI’s visual builder.

Agents in Handit.AI are built as graphs made up of nodes. Each node represents a tool or a model, and you define how they connect and interact. This step sets the foundation for how your agent will behave and respond in real time.

  1. Log in to your Handit.AI Dashboard.

  2. In the sidebar, click "Your Agents".

  3. Click "Create Agent" to start building a new agent.


Step 1: Add Nodes

  • Click "Add Node" for each step in your agent’s workflow.

  • Choose a node type:

    • Model – for prediction/classification/generation

    • Tool – for actions like API calls or custom logic

  • Fill in the required configuration for each node (e.g., model name, inputs, parameters).


Step 2: Connect Nodes

  • Drag edges between nodes to define the flow of your agent.

  • Each connection determines how data moves from one node to the next.


Step 3: Finalize Agent

  • Click "Create" once your graph is ready.

  • Fill in your agent details:

    • Name – A clear and unique name for your agent

    • Description – A short summary of what your agent does

2. Download and Set Up MCP Server (1 minute)

To connect your local code to the agent you just created, you'll need the MCP (Model Context Protocol) server.

This lightweight server acts as a bridge between Cursor and your Handit.AI agent. Once configured, it enables deep integration and real-time monitoring inside your development workflow.

  1. Download your MCP Server zip from Handit

  2. Extract the MCP server zip file to your project directory

  3. Open Cursor Settings

  4. Go to Features > MCP

  5. You should see the Handit.AI MCP server

  6. Click "Enable"

3. Connect Your Code (1 minute)

Now it’s time to integrate your actual code with the agent you built.

With the MCP server enabled, you can use Cursor’s AI assistant to automatically instrument your code for agent monitoring and tracing.

  1. Use the agent input in Cursor

  2. Type "configure Handit on my agent on these files .... Based on the MCP handit://context"

  3. Review the suggested changes

  4. Accept the changes if everything looks correct

What You Get

Once everything is connected, Handit.AI gives you full visibility and optimization over your agent’s behavior.

  • ✅ Live Traces: See every step your agent takes

  • ✅ Performance Metrics: Track accuracy and efficiency

  • ✅ Self-Optimization: Watch your agent improve automatically (after collecting traces)

Need Help?

If something seems wrong, check the

Check out our for detailed agent setup

Visit our for server configuration

Explore for optimization

Manual Setup Guide
Frontend Creation Guide
MCP Server Setup Guide
Best Practices