---
title: Connect Datadog
url: /docs/ai/agents/integrations/datadog/
---


Connect Datadog so an agent can query monitoring data during an investigation. Neo and agents you configure share the organization's connection, which uses a Datadog API key and application key.

## Prepare Datadog access

In Datadog **Organization Settings**, open **Service Accounts** and create or select the account. Enable **MCP Access** in the organization settings. Give the service account **MCP Read** (`mcp_read`) and the resource-level permissions for the data the agent needs. Limit its application key to the required scopes. For an investigation, omit permissions to change monitors unless that is part of the job. See [Design access around the job](/docs/ai/agents/permissions/#design-access-around-the-job) for shared access guidance.

## Create the credentials

1. Create a scoped application key for that service account.
1. Open **Organization Settings** > **API Keys** and create an API key in the same organization.
1. Identify your Datadog site: `us1`, `us3`, `us5`, `eu1`, `ap1`, or `ap2`.

See [Datadog MCP authentication](https://docs.datadoghq.com/mcp_server/setup/#api-and-application-keys) for the key requirements.

## Connect Datadog

With permission to manage organization integrations (`org_integrations:update`):

1. Open **Agent Settings** > **Integrations** > **MCP servers**.
1. Find **Datadog** under **Available** and select **Authorize**.
1. Enter **Datadog site**, **API Key**, and **App Key**, then select **Save**.

For an agent you configure, [select Datadog in its configuration](/docs/ai/agents/configure/#connect-the-tools-it-needs). For Neo, Datadog is available to new tasks by default. Turn it off in the task composer if a task doesn't need it.