Prerequisites:

​
1. Download the Open Agent Platform code

1

Clone the Repository

Clone the Open Agent Platform Repository from GitHub
2

Set Environment Variables

The repository contains a .env.example file listing all of the environment variables you need to run the platform.
Set the following environment variables:
  NEXT_PUBLIC_BASE_API_URL="http://localhost:3000/api"
  LANGSMITH_API_KEY="lsv2_..."
  # Or whichever LLM's API key you're using
  OPENAI_API_KEY="..."

​
2. Authentication Setup

Open Agent Platform uses Supabase for authentication by default.
1

Create a Supabase Project

Create a new project in Supabase.
2

Configure Environment Variables

Set the following environment variables inside the apps/web/ directory:
You can find these Next.js-specific variables in the “Connect” window in the Supabase dashboard, under “App Frameworks”
NEXT_PUBLIC_SUPABASE_URL="<your supabase url>"
NEXT_PUBLIC_SUPABASE_ANON_KEY="<your supabase anon key>"
3

Enable Authentication Providers

Enable Google authentication in your Supabase project, or set NEXT_PUBLIC_GOOGLE_AUTH_DISABLED=true in your environment variables to disable showing Google as an authentication option in the UI.

​
3. Deploying Agents

The next step in setting up Open Agent Platform is to deploy and configure your agents.
1

Clone Pre-built Agents

We’ve released three pre-built agents specifically for Open Agent Platform:
2

Follow Agent Setup Instructions

For each agent repository:
  1. Clone the repository
  2. Follow the instructions in the README
  3. Deploy the agents to LangGraph Platform, or run langgraph dev to run the agents locally. Optionally pass --port <port> to use a custom port. This is useful if running multiple graphs locally.
3

Configure Environment Variables

After deployment, create a configuration object for each agent (for the next step):
{
  "id": "The project ID of the deployment",
  "tenantId": "The tenant ID of your LangSmith account",
  "deploymentUrl": "The API URL to your deployment",
  "name": "A custom name for your deployment",
  "isDefault": "Whether this deployment is the default deployment (only one can be default)",
  "defaultGraphId": "The graph ID of the default graph (optional, only required if isDefault is true)"
}
You can find your project & tenant IDs with a GET request to the /info endpoint on your LangGraph Platform deployment.
If you are running agents locally via langgraph dev, the id (project ID), and tenantId should be any valid UUID version 4, such as those generated by uuid.uuid4(). Ensure each graph has a unique id, and all graphs share the same tenantId.
4

Set Environment Variables

Combine your agent configurations into a JSON array and set the NEXT_PUBLIC_DEPLOYMENTS environment variable inside the apps/web/ directory:
NEXT_PUBLIC_DEPLOYMENTS=[{"id":"bf63dc89-1de7-4a65-8336-af9ecda479d6","deploymentUrl":"http://localhost:2024","tenantId":"42d732b3-1324-4226-9fe9-513044dceb58","name":"Local deployment","isDefault":true,"defaultGraphId":"agent"}]

​
4. RAG Server Setup

1

Deploy LangConnect

Follow the instructions in the LangConnect README to set up and deploy a LangConnect RAG server.
2

Configure Environment Variables

Set the RAG server URL inside the apps/web/ directory:
NEXT_PUBLIC_RAG_API_URL="http://localhost:8080"

​
5. MCP Server Setup

Open Agent Platform only supports connecting to MCP servers which support Streamable HTTP requests. If you don’t have an MCP server set up yet, you can use Arcade’s public beta MCP server.
1

Configure MCP Server URL

Set your MCP server URL inside the apps/web/ directory (ensure it does not end with /mcp. This will be auto-applied by OAP):
NEXT_PUBLIC_MCP_SERVER_URL="https://api.arcade.dev/v1/mcps/arcade-anon/"
2

Configure Authentication (if required)

For authenticated MCP servers:
NEXT_PUBLIC_MCP_AUTH_REQUIRED=true

​
6. Run Your Platform

Start the application with your configured environment variables: 
# Navigate to the web app directory
cd apps/web

# Install dependencies
yarn install

# Start the development server
yarn dev
Your Open Agent Platform should now be running at http://localhost:3000!