Arlet
Datadog

Use Datadog from your AI

Search Datadog logs, metrics, traces, monitors, incidents, dashboards and RUM events, and create notebooks through Datadog's official MCP server (core toolset). Connect by signing in.

Enable Datadog in ArletDatadog official docs

Arlet makes Datadog's official MCP server available from the AI client you already use. Signing in to Datadog completes the connection.

The AI can search logs, metrics, traces, monitors, incidents, dashboards and RUM events and create notebooks (core toolset). This targets the US1 site (datadoghq.com), and your role needs the mcp_read permission (mcp_write for writes). The available tools are those provided by Datadog's official server.

Choose your AI client

Pick the AI client you use to see how to connect it with Datadog.

What you can do

Tools the AI can call through the Datadog integration.

  • search_datadog_logs

    Search logs.

  • analyze_datadog_logs

    Aggregate and analyze logs.

  • search_datadog_metrics

    Search metrics.

  • get_datadog_metric

    Get metric values.

  • search_datadog_spans

    Search APM spans.

  • get_datadog_trace

    Get a trace.

  • search_datadog_monitors

    Search monitors.

  • search_datadog_incidents

    Search incidents.

  • get_datadog_incident

    Get an incident's details.

  • search_datadog_events

    Search events.

  • search_datadog_hosts

    Search hosts.

  • search_datadog_dashboards

    Search dashboards.

  • search_datadog_rum_events

    Search RUM events.

  • search_datadog_notebooks

    Search notebooks.

  • create_datadog_notebook

    Create a notebook.

The tool list follows what Datadog's official MCP server provides and may change.

Example uses

  • Investigate an outage

    Start from the firing monitors, follow the related logs and traces, and list likely causes.

  • Understand error trends

    Aggregate recent error logs by service and error type.

  • Write a postmortem

    Collect an incident's timeline and related metrics into a notebook.

  • Check performance

    Query latency metrics and spans to find the slowest endpoints.

Prompts you can use as-is

  • List the monitors currently alerting and show five related error logs for each
  • Aggregate the last hour of error logs for the checkout service by error type
  • Collect the timeline and related metrics for incident #123 and create a postmortem notebook
  • Show the five endpoints in the api service with the highest p95 latency