New Relic MCP Server
Query New Relic telemetry with natural language or NRQL, investigate alerts and incidents, and analyze service performance directly from AI clients.
The New Relic MCP Server is an official, provider-hosted remote MCP server that connects AI agents to the New Relic observability platform. It exposes telemetry, entities, dashboards, alerts, incidents, and logs through a standardized MCP interface so assistants like Claude, Cursor, VS Code Copilot, Claude Code, and Gemini CLI can investigate production issues using natural language. The server is currently in public preview.
The server groups tools into six functional areas: entity and account management (entity lookup, dashboards, accounts), data access (NRQL execution and natural language to NRQL translation), alerting and monitoring (policies, conditions, recent issues, synthetic monitors), incident response (change events, error groups, user impact reports), performance analytics (golden metrics, transactions, threads, GC, Kafka, logs), and advanced analysis (deployment impact, alert insights). Authentication is supported via OAuth 2.0 (browser-based login) or a User API key in NRAK- format.
The server is hosted by New Relic at https://mcp.newrelic.com/mcp/ (US) and https://mcp.eu.newrelic.com/mcp/ (EU). The repo at github.com/newrelic/mcp-server is the reference, but there is no self-hosted binary; clients connect to the hosted endpoint via the mcp-remote proxy or a native HTTP MCP transport. Use with FedRAMP or HIPAA-regulated New Relic accounts is prohibited during public preview.
Tools
| Tool | Description |
|---|---|
execute_nrql_query |
Execute an NRQL query against NRDB. |
natural_language_to_nrql_query |
Convert a natural language request into an NRQL query, execute it, and return results. |
get_entity |
Fetch New Relic entities by GUID or search by name pattern. |
list_related_entities |
List entities one hop away from a given entity GUID. |
list_entity_types |
List the complete catalog of New Relic entity types with domain/type definitions. |
search_entity_with_tag |
Search for entities using a specific tag key and value. |
list_available_new_relic_accounts |
List all account IDs available to the user. |
get_dashboard |
Retrieve details for a specific dashboard. |
list_dashboards |
List all dashboards for a New Relic account. |
convert_time_period_to_epoch_ms |
Convert a time period like 'last 30 minutes' to epoch milliseconds. |
list_alert_policies |
Query alert policies with optional name filtering. |
list_alert_conditions |
List alert condition details for a specific alert policy. |
search_incident |
List alert events (open and close) with flexible filtering. |
list_recent_issues |
List all open issues in New Relic for the specified account. |
list_synthetic_monitors |
List all synthetic monitors that check service availability and performance. |
list_change_events |
List a history of change events from New Relic for the application. |
list_entity_error_groups |
Fetch error groups for a specific entity from the Errors Inbox within a time window. |
analyze_deployment_impact |
Evaluate performance effects of recent deployments. |
generate_alert_insights_report |
Generate an intelligence report for alerts. |
generate_user_impact_report |
Generate an end-user impact analysis report for a specific issue. |
analyze_golden_metrics |
Evaluate key health indicators including throughput, latency, and error rate. |
analyze_transactions |
Analyze transactions for a specific entity within a time window. |
analyze_threads |
Analyze thread metric data including state, CPU usage, and memory consumption. |
analyze_kafka_metrics |
Analyze Kafka metrics including consumer lag, producer throughput, latency, and partition balance. |
list_garbage_collection_metrics |
Retrieve memory and garbage collection metrics for an entity. |
analyze_entity_logs |
Analyze application logs to identify error patterns, anomalies, and recurring issues. |
list_recent_logs |
List recent logs from New Relic for the specified account and entity GUID. |
Prerequisites
- A New Relic account (US or EU region) with API access
- A User API key in
NRAK-XXXXXXXXXXXXXXXXXXXXXformat (created via the New Relic UI under API keys), or use OAuth login - Node.js installed (required for the
mcp-remoteproxy used by Claude Desktop) - Membership in a New Relic group with appropriate organizational role permissions
- Public preview only: do not use with FedRAMP or HIPAA-regulated accounts
Server endpoints
- US region:
https://mcp.newrelic.com/mcp/ - EU region:
https://mcp.eu.newrelic.com/mcp/
Claude Desktop, OAuth (recommended)
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"new-relic-mcp": {
"command": "npx",
"args": ["mcp-remote", "https://mcp.newrelic.com/mcp/"]
}
}
}
A browser window will open for OAuth login on first use.
Claude Desktop, API key (macOS)
{
"mcpServers": {
"newrelic": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.newrelic.com/mcp/", "--transport", "http", "--header", "api-key: NRAK-xxxx"]
}
}
}
On Windows, use C:\\PROGRA~1\\nodejs\\npx.cmd as the command value.
VS Code
Create .vscode/mcp.json in your project root.
OAuth:
{
"servers": {
"new-relic-mcp": {
"url": "https://mcp.newrelic.com/mcp/",
"type": "http"
}
}
}
API key:
{
"servers": {
"new-relic-mcp": {
"url": "https://mcp.newrelic.com/mcp/",
"headers": {
"api-key": "NRAK-YOUR-API-KEY-HERE"
}
}
}
}
Other supported clients listed in the docs include Cursor, Windsurf, Claude Code, Gemini CLI, and Kiro CLI. For EU accounts, replace the host with mcp.eu.newrelic.com.
- Ask plain-English questions about service health and let the agent translate them into NRQL via
natural_language_to_nrql_query, no NRQL fluency required. - Triage active incidents during an on-call shift by combining
list_recent_issues,list_change_events, andanalyze_deployment_impactto correlate alerts with recent deploys. - Debug a slow endpoint by chaining
get_entity,analyze_golden_metrics,analyze_transactions, andanalyze_threadsfor a target service. - Investigate error spikes with
list_entity_error_groupsandanalyze_entity_logsto find recurring exceptions and impacted users. - Generate post-incident summaries using
generate_alert_insights_reportandgenerate_user_impact_reportfor a given issue.
- "Run an NRQL query to show p95 latency for the checkout service over the last 6 hours, grouped by host."
- "What are the open issues in my production account right now, and which deployments happened in the last 24 hours on the affected services?"
- "Analyze the golden metrics and thread state for the
payments-apientity for the past hour and tell me what changed." - "Pull the most recent error groups for the
web-frontendservice from the Errors Inbox and summarize the top three by user impact." - "Generate a user impact report for issue ID 12345 and list which deployments may have caused it."
- Official, provider-hosted server maintained by New Relic, so no local install or binary to manage.
- Broad tool coverage across NRQL, entities, dashboards, alerts, incidents, logs, deployments, and performance analytics.
- Supports both OAuth 2.0 and User API key authentication, with US and EU endpoints.
- Includes a natural language to NRQL tool, lowering the barrier for users unfamiliar with NRQL syntax.
- Currently in public preview, with restrictions including no use on FedRAMP or HIPAA-regulated accounts.
- Requires Node.js and the
mcp-remoteproxy for OAuth from Claude Desktop, adding a moving part to the setup. - Tool input schemas and rate limits are not fully documented in the public README; details live in the New Relic docs portal.
- Datadog MCP server for similar observability workflows on Datadog.
- Grafana MCP server for Grafana dashboards, Loki logs, and Prometheus queries.
- Community New Relic MCP implementations such as
cloudbring/newrelic-mcpandulucaydin/mcp-server-newrelicfor self-hosted alternatives.