Overview

The Datadog MCP Server uses the Model Context Protocol (MCP) to provide AI agents with access to Feature Flags management capabilities, including flag creation, configuration, and React/JavaScript integration guidance.

Setup

The following configurations give your AI agent access to the Feature Flags toolset in the Datadog MCP Server. You must restart your agent after performing this setup.

For all clients, install the MCP Server binary:

curl -sSL https://coterm.datadoghq.com/mcp-cli/install.sh | bash

Then follow the instructions below to add the MCP Server to your specific client.

Claude Code

claude mcp add datadog -- ~/.local/bin/datadog_mcp_cli --endpoint-path /api/unstable/mcp-server/mcp?toolsets=feature-flags

Cursor

Add this to ~/.cursor/mcp.json (remember to save the file):

{
  "mcpServers": {
    "datadog-ff": {
      "type": "stdio",
      "command": "~/.local/bin/datadog_mcp_cli --endpoint-path /api/unstable/mcp-server/mcp?toolsets=feature-flags",
      "args": [],
      "env": {}
    }
  }
}

Use cases

The MCP Server includes tools to help you manage feature flags in your codebase. The following use cases provide sample prompts for using the tools.

Create feature flags

Use the create-feature-flag tool to create feature flags. You do not need to specify the tool name in the prompt, but it can provide more consistent results. The MCP Server has access to Datadog’s documentation and uses it to implement the flag in your codebase.

If you do not yet have feature flags implemented, mention in the prompt that you want to implement Datadog feature flags.

Example prompts:

• Use the create-feature-flag tool to create a flag to control the title on the main page.
• I want to show a confirmation modal when <SOME_EVENT> happens. Use a Datadog feature flag to control whether the confirmation modal is shown.

Check feature flag implementation

Use the check-flag-implementation tool to check if a feature flag is implemented correctly.

The tool checks that the flag is being referenced as the correct value type, is passing the correct subject attributes, and is providing the correct default value that agrees with the default in production environments.

Example prompts:

• Check if the show-confirmation-modal flag is implemented correctly.
• Check if all feature flags in /some/directory are implemented correctly.

Note: This may not find all issues. Checking flags individually is more reliable.

This tool can also be used to add feature flags created in the UI to your codebase. For example:

• Use the show-confirmation-modal flag to control whether the confirmation modal is shown when <SOME_EVENT> happens.

List feature flags

Use the list-feature-flags tool to list all feature flags. For example:

• List all feature flags.

List environments

Use the list-environments tool to list all environments. For example:

• List my flagging environments.

Update feature flag environments

Use the update-feature-flag-environment tool to update a feature flag environment. This tool can control the default variants, and enable or disable the flag. It cannot modify flags in production environments.

Example prompts:

• I want show-confirmation-modal to serve true in development.
• Disable show-confirmation-modal in staging.

Further reading

Additional helpful documentation, links, and articles:

Getting Started with Feature FlagsDOCUMENTATION Datadog MCP ServerDOCUMENTATION