LaunchDarkly's Model Context Protocol (MCP) Server

Name: LaunchDarkly's Model Context Protocol (MCP) Server
Author: launchdarkly

@launchdarkly

Repository

About LaunchDarkly's Model Context Protocol (MCP) Server

#featureflags #configuration #devtools

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "LaunchDarkly": {
      "command": "npx",
      "args": [
        "-y",
        "--package",
        "@launchdarkly/mcp-server",
        "--",
        "mcp",
        "start",
        "--api-key",
        "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is LaunchDarkly's Model Context Protocol (MCP) Server?

A self-managed MCP server that enables AI clients to interact with LaunchDarkly’s feature flags, AI configs, audit logs, environments, and code references. It is designed for LaunchDarkly Federal and European Union (EU) instances where the hosted MCP server is not yet available.

How to use LaunchDarkly's Model Context Protocol (MCP) Server?

Install via npx, Docker, standalone binary, or local clone. Configure your AI client (Cursor, Claude, Qodo Gen, GitHub Copilot CLI) with the mcp start command and a LaunchDarkly API key. For EU or Federal instances, add the --server-url argument with the appropriate environment URL.

Key features of LaunchDarkly's Model Context Protocol (MCP) Server

List, create, update, and delete feature flags
Manage AI Configs and their targeting
Show and update AI Config variations
List audit log entries
List environments by project
List code reference repositories
Support for commercial, Federal, and EU environments

Use cases of LaunchDarkly's Model Context Protocol (MCP) Server

Enable an AI coding assistant to create or modify feature flags during development
Allow an AI agent to audit flag changes by listing recent audit log entries
Manage AI Configs and their variations from a chat interface
Query environment metadata across projects directly from your AI client

FAQ from LaunchDarkly's Model Context Protocol (MCP) Server

How is this different from LaunchDarkly’s hosted MCP server?

The hosted MCP server is more feature complete and receives more frequent updates. LaunchDarkly strongly recommends using the hosted server where available. This self-managed server exists for LaunchDarkly EU or Federal instances that do not yet support the hosted version.

What dependencies are required?

For JavaScript runtimes, see RUNTIMES.md. The standalone binary has no additional dependencies. For npx installation, Node.js and npm must be available.

Where does flag and config data live?

Data remains in LaunchDarkly. The MCP server connects to the LaunchDarkly API using your API key. No customer data is stored locally by the server.

What transport or authentication does the server use?

The server uses the MCP protocol. Authentication is via a LaunchDarkly API key passed as a --api-key argument. For Docker, the server listens on localhost:8080 with SSE transport.

Are there any known limitations?

The self-managed server is less feature complete than the hosted version and receives fewer updates. No other explicit limits are documented in the README.