OfficialVerified

LaunchDarkly MCP Server

LaunchDarkly feature flag management for EU and Federal instances via MCP.

Data last scanned today · Reviewed today

33 1 dangerous tool7 tools★ 27TypeScriptMIT

Overview

This server provides a local Model Context Protocol integration for LaunchDarkly environments specifically located in the European Union or those within Federal instances. It serves as a self-managed bridge to facilitate interaction with feature flags in environments where hosted solutions are not yet available. To use this server, you must configure your AI client with a valid LaunchDarkly API key. The implementation is designed to be installed as a local package via npx, allowing AI agents to interface with your existing feature management infrastructure.

Our verdict

With a security score of 40/100, this server is currently classified as high risk. Because it is community-maintained rather than an official vendor release and has not undergone editorial review, users should exercise caution. LaunchDarkly explicitly recommends using their hosted MCP server whenever possible, limiting this self-managed version to users requiring access to EU or Federal instances that lack hosted support.

Supports specialized EU and Federal environment configurations.
Low supply-chain risk with only two direct dependencies.
Active development with a recent commit history.

Not an official vendor-maintained implementation.
Lacks an editorial security review.
Authentication methods have not yet been security vetted.

Setup

Claude Desktop

{
  "mcpServers": {
    "launchdarkly-mcp-server": {
      "args": [
        "-y",
        "@launchdarkly/mcp-server"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
getStatus	Get flag status across environments	low
list	List feature flags	low
create	Create a feature flag	medium
get	Get feature flag	low
patch	Update feature flag	medium
delete	Delete feature flag	high
getFeatureFlags	Retrieve a list of feature flags for a given project key.	low

Compatibility

Client	Local	Docker	Remote	Read-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›When should I use this local MCP server instead of the hosted version?

You should use this self-managed server specifically if you are using LaunchDarkly Federal or European Union (EU) instances, which do not currently support the hosted MCP server.

›What is required to authenticate the LaunchDarkly MCP server?

You must provide a LaunchDarkly API key, which can be found on your LaunchDarkly Authorization page.

›Can I run the MCP server without installing additional dependencies like Node.js?

Yes, you can run the server as a standalone binary by downloading the appropriate file from the GitHub releases page.

›Is it possible to manage the API key using environment variables?

Yes, for AI clients that support environment variable access in their MCP configuration, you can map an environment variable like LD_ACCESS_TOKEN to the --api-key argument.

›How can I run the server if I want to modify the source code locally?

You can clone the repository, run 'npm install' and 'npm run build', and then configure your MCP server definition to execute the local 'mcp-server.js' file using node.