LaunchDarkly local MCP server
The local Model Context Protocol (MCP) server for LaunchDarkly federal and European Union (EU) environments.
</a>
[!IMPORTANT]
接入你的智能体
發布者自己的指令。不是 PlugMCP 端點。
用我們的金鑰與日誌託管一份副本。
npx -y @launchdarkly/[email protected]Claude Code 沒有本機指令的安裝連結。複製下方指令並在終端機中執行。
claude mcp add launchdarkly-mcp-server -- npx -y @launchdarkly/mcp-server@0.7.2在終端機中執行。
{ "mcpServers": { "launchdarkly-mcp-server": { "command": "npx", "args": [ "-y", "@launchdarkly/[email protected]" ] } }}claude_desktop_config.json
遠端 URL 還是本機指令? 兩種安裝方式的差異
來自發布者
The local Model Context Protocol (MCP) server for LaunchDarkly federal and European Union (EU) environments.
</a>
[!IMPORTANT]
The local [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server for [LaunchDarkly](https://launchdarkly.com/) federal and European Union (EU) environments.
LaunchDarkly strongly recommends using the hosted MCP server where it is available. The hosted server is more feature complete and receives more frequent updates than this self-managed server.
LaunchDarkly provides this self-managed MCP server for LaunchDarkly EU or Federal instances, which do not yet support the hosted MCP.
This MCP server can be installed in any AI client that supports the MCP protocol. Refer to your AI client's instructions if it isn't listed here.
[!IMPORTANT] Some eMudhra intermediate certificate authorities were compromised in October 2026 (advisory). Anyone on your network who holds one of their keys could intercept the server's connection to LaunchDarkly and read your API access token.
- npx and local installs: use Node.js 22.23.4, 24.22.1, 26.12.1 or later. These releases reject the compromised intermediates. The server also rejects them itself and prints a warning on startup when your runtime does not.
- Docker: use image 0.7.1 or later. Those run on Node.js 24.22.1; 0.7.0 and earlier run on Node.js 24.2.0.
- Standalone binaries: stop using the
mcp-server-bun-*binaries attached to releases v0.6.1 and earlier. They embed a Bun runtime that does not reject the compromised intermediates, and they are no longer published. Switch to npx with a patched Node.js or to Docker.
Create a .cursor/mcp.json file in your project root with the following content:
{
"mcpServers": {
"LaunchDarkly": {
"command": "npx",
"args": [
"-y", "--package", "@launchdarkly/mcp-server", "--", "mcp", "start",
"--api-key", "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
]
}
}
}
Specify your API key as found on LaunchDarkly's Authorization page.
Add the following server definition to your claude_desktop_config.json file:
{
"mcpServers": {
"LaunchDarkly": {
"command": "npx",
"args": [
"-y", "--package", "@launchdarkly/mcp-server", "--", "mcp", "start",
"--api-key", "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
]
}
}
}
Specify your API key as found on LaunchDarkly's Authorization page.
Connect more tools.+ Add new MCP.{
"mcpServers": {
"LaunchDarkly": {
"command": "npx",
"args": [
"-y", "--package", "@launchdarkly/mcp-server", "--", "mcp", "start",
"--api-key", "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
]
}
}
}
Specify your API key as found on LaunchDarkly's Authorization page.
Save.Use the Copilot CLI to interactively add the MCP server:
/mcp add
Or create/edit ~/.copilot/mcp-config.json with the following content:
{
"mcpServers": {
"LaunchDarkly": {
"command": "npx",
"args": [
"-y", "--package", "@launchdarkly/mcp-server", "--", "mcp", "start",
"--api-key", "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
]
}
}
}
Specify your API key as found on LaunchDarkly's Authorization page.
For more information, see the GitHub Copilot CLI documentation.
You can also run the MCP server locally by cloning this repository. Once cloned, you'll need to install dependencies (npm install) and build the server (npm run build).
Then, configure your server definition to reference your local clone. For example:
{
"mcpServers": {
"launchdarkly": {
"command": "node",
"args": [
"/path/to/mcp-server/bin/mcp-server.js", "start",
"--api-key", "api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
]
}
}
}
Some AI clients allow for accessing environment variables within MCP configurations. In AI clients where this is supported, you can configure your MCP server to read from an environment variable like MCP_LD_TOKEN as follows:
{
"mcpServers": {
"launchdarkly": {
"command": "npx",
"args": [
"-y", "--package", "@launchdarkly/mcp-server", "--", "mcp", "start",
"--api-key", "$LD_ACCESS_TOKEN"
],
"env": {
"LD_ACCESS_TOKEN": "MCP_LD_TOKEN"
}
}
}
}
If installing the MCP server from the AWS Markeplace, pull the image using an authenticated role using the provided instructions in the marketplace listing.
To run the container:
docker run --rm -p 127.0.0.1:8080:8080 709825985650.dkr.ecr.us-east-1.amazonaws.com/launchdarkly/mcp --api-key api-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx --auth-token "$MCP_AUTH_TOKEN"
The container listens on 0.0.0.0, so --auth-token (at least 16 characters) is required. Clients must send Authorization: Bearer <token> on /sse and /message. Outside Docker, the SSE transport listens on 127.0.0.1 by default; pass --host to change it, which also requires --auth-token for non-loopback hosts.
Then, configure your server definition to point at the running container. For example:
{
"mcpServers": {
"launchdarkly": {
"url": "http://localhost:8080/sse",
"headers": {
"Authorization": "Bearer <your --auth-token value>"
}
}
}
}
This won't work in AI clients (such as Cursor) which don't support accessing environment variables directly within MCP configurations. As a workaround, you can invoke a local script from your MCP configuration. See here for an example.
For supported JavaScript runtimes, please consult RUNTIMES.md.
Most customer accounts run on LaunchDarkly's commercial (default) environment. Customers on other environments can specify the --server-url argument to connect to the appropriate environment. For example, customers on LaunchDarkly's Federal environment should specify the --server-url https://app.launchdarkly.us argument when starting their MCP server.
While we value open-source contributions to this SDK, this library is generated programmatically. Any manual changes added to internal files will be overwritten on the next generation. We look forward to hearing your feedback. Feel free to open a PR or an issue with a proof of concept and we'll do our best to include it in a future release.
The local [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server for [LaunchDarkly](https://launchdarkly.com/) federal and European Union (EU) environments.
{ "mcpServers": { "launchdarkly-mcp-server": { "command": "npx", "args": [ "-y", "@launchdarkly/[email protected]" ] } }}工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
這筆記錄未列出必要的環境變數。
@launchdarkly/ 透過 stdio 在本機執行。
上次檢查時間:2026-11-16。 The local [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server for [LaunchDarkly](https://launchdarkly.com/) federal and European Union (EU) environments.