New Relic Ground Truth

Ground Truth is New Relic's operational intelligence layer. Model Context Protocol (MCP) tools help organize your observability data so that AI agents can better understand your tech stack.

MCP is a standard way for AI models or agents to receive relevant information (context) about a specific system or environment. This context can help the AI understand the current situation, perform tasks more accurately, and provide more useful responses. The MCP server allows supported AI agent(s) to call certain tools that may help you query, investigate, or act on New Relic data in natural language. This in turn can help speed incident response and performance analysis.

Potential benefits

The MCP server offers practical benefits for engineers and operations teams in the following ways:

Direct integration: Access New Relic's MCP tools directly from your supported development environment. This avoids switching between different AI tools.
Faster troubleshooting: Use MCP to help quickly find and fix issues from your observability data, analyze error logs, and determine probable causes with AI-provided insights.
Operational queries: Ask your AI agent to check alert statuses, generate alert event reports, or analyze how deployments affect your systems.
シンプルなデータクエリ：New Relicクエリ言語（NRQL）に詳しくなくても、平易な英語で質問を作成すると、サーバーがそれをNRQLに変換して結果を取得できます。

サポート対象のAI開発ツール

New Relic MCP works with these AI development environments (supported MCP clients):

Claude Code：Claude向けのコマンドラインインターフェース
Antigravity CLI: Command-line interface for Antigravity
Kiro CLI：Kiro向けのコマンドラインインターフェース
Claude Desktop：インタラクティブなAI開発のためのデスクトップアプリケーション
VS Code：MCPに対応している統合開発環境

Each environment offers unique advantages depending on your workflow preferences. Refer to our setup guide for MCP client-specific configuration instructions.

前提条件

New Relic MCPを使用するには、以下が必要です。

A New Relic account (MCP is available to all accounts: all data and CCU users can access core tools, aCCU users can access all tools including advanced tools)
Node.js のインストール（一部の認証方法で必要）
One of the supported MCP clients mentioned earlier installed

Organization access

Before you can connect, a New Relic administrator must perform the following:

グループメンバーシップとロール

You must be a member of a group assigned either a default organizational role (Organization Read Only, Organization Manager, or Organization Product Admin) or a custom organization-scoped role with permission to read the MCP server. If your group does not have the required permission, contact an administrator with the Organization Manager or Authentication Domain Manager role to have it granted.

Feature control manager

Your organization administrator must enable MCP access via feature control manager. The following toggles control MCP access. New Relic MCP Server and Local Clients are enabled by default. All other sub-toggles are disabled by default and can be enabled by your New Relic administrator as needed.

New Relic MCP Server: Required for MCP set up. Main toggle that must be on for any MCP access to work.
Local Clients: Required for local client connections (Claude Code, Claude Desktop, VS Code, Kiro).
API Keys: Required for API key connections, including automated scripts.
Preview Tools: Required if you want to access MCP preview (non-GA) tools.
Atlassian: Required for Atlassian integration
AWS: Required for AWS integration

認証オプション

OAuth (recommended): New Relic provides pre-configured OAuth endpoints:
- クライアントID： pUWGgnjsQ0bydqCbavTPpw==
- 認証URL： https://oauth2.service.newrelic.com/oauth2/auth
- トークンURL： https://oauth2.service.newrelic.com/oauth2/token
- スコープ： ["observability:read", "offline", "offline_access"]
User API Key: Note that some MCP tools are only accessible via OAuth.
1. New Relicにログインしてください
2. ユーザーメニュー -> API keysを選択します。
3. 新しいユーザーキーを作成するか、既存のユーザーキーをコピーします。
4. キーの形式：NRAK-XXXXXXXXXXXXXXXXXXXXX。

重要

Some tools are only accessible via OAuth and are not available when using an API key. These include AI-powered analysis tools (natural_language_to_nrql_query, generate_alert_insights_report, generate_user_impact_report, analyze_deployment_impact) and preview tools (search_traces, analyze_trace_anomaly, analyze_service_dependencies). Refer to the tool reference for details.

次のステップ

Ready to connect your supported MCP clients to New Relic observability data?

Choose your supported MCP client from our supported environments.
設定手順については、セットアップガイドをご覧ください。
利用可能なツールについては、ツールリファレンスをご覧ください。
自然言語を使ってNew Relicのデータを照会できます。
一般的な問題のトラブルシューティングを行います。