home / mcp / devcycle cli & mcp server

DevCycle CLI & MCP Server

Provides a CLI for managing DevCycle feature flags and an MCP server for AI assistants to control feature flags.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "devcyclehq-cli": {
      "url": "https://mcp.devcycle.com/mcp",
      "headers": {
        "DEVCYCLE_CLIENT_ID": "<DEVCYCLE_CLIENT_ID>",
        "DEVCYCLE_PROJECT_KEY": "<DEVCYCLE_PROJECT_KEY>",
        "DEVCYCLE_CLIENT_SECRET": "<DEVCYCLE_CLIENT_SECRET>"
      }
    }
  }
}

You can use the DevCycle MCP Server to control feature flags directly from AI coding assistants or your development workflow. This server enables AI-powered interactions for creating, updating, and managing features, variables, and targeting rules in DevCycle, making it easier to integrate feature management into your coding environment.

How to use

The MCP server lets your AI assistants manage DevCycle data sources and actions from inside your development environment. You can connect via a hosted MCP URL or run a local MCP client that communicates with the server, then your AI assistant can create, update, and manage feature flags, variables, and related configurations.

How to install

Prerequisites: you should have Node.js and npm installed on your machine. If you plan to use the hosted MCP server, no local installation is required for MCP. If you prefer a local setup, follow the steps below to run the MCP client that connects to the hosted server.

# Ensure Node.js is installed
node -v
npm -v

# If you want to run a local MCP client that connects to the hosted server
# Install the MCP client globally (example shown for npm-based setup)
npm install -g @devcycle/cli

# Alternatively, if you want to run the MCP client via npx directly
npx mcp-remote https://mcp.devcycle.com/mcp

Configuration and connection details

Configure your AI assistant to connect to the hosted MCP server. The hosted endpoint is https://mcp.devcycle.com/mcp. You can add this MCP URL to your AI assistant configuration so it can authenticate and start managing feature flags.

If you are using Cursor, add the MCP server to your cursor settings as follows.

{
  "mcpServers": {
    "devcycle": {
      "url": "https://mcp.devcycle.com/mcp"
    }
  }
}

If you are using Claude Desktop, configure your Claude settings to point to the hosted MCP server with the local wrapper command.

{
  "mcpServers": {
    "devcycle": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.devcycle.com/mcp"
      ]
    }
  }
}

Hosted MCP setup flow

Using the hosted MCP server, you will be guided through OAuth authentication the first time you connect. This enables a secure, token-based interaction between your AI assistant and DevCycle.

Advanced usage notes

Your AI assistant can create, update, and manage feature flags, variables, variations, and targeting rules once connected. You can also leverage the MCP server in workflows that require real-time flag management directly from your development environment.

Troubleshooting and tips

If authentication fails in a containerized environment, ensure that the required ports are accessible for the OAuth flow (for example, 2194, 2195, 2196, or 8080) so the CLI can receive the token correctly.

Examples and quick-start snippets

{
  "mcpServers": {
    "devcycle": {
      "url": "https://mcp.devcycle.com/mcp"
    }
  }
}
```

```json
{
  "mcpServers": {
    "devcycle": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.devcycle.com/mcp"]
    }
  }
}