---
title: "Let your AI agent set up Datadog with Agentic Onboarding"
description: "Learn how Agentic Onboarding simplifies Datadog setup across applications and infrastructure through the Pup CLI or an MCP-compatible coding assistant."
author: "Vidur Khanna, April Yep"
date: 2026-11-20
tags: ["infrastructure monitoring", "real user monitoring", "ai"]
blog_type_id: the-monitor
locale: en
---

Developers use AI coding agents for a variety of software development tasks, from implementing features to updating deployment configurations. But setting up visibility into that software can remain a separate process. After an application is ready, developers still need to find the appropriate observability setup instructions, determine which configuration matches their project, and install dependencies. They must then translate general setup documentation into changes that fit their stack across application code and deployment files. As more development work moves into agentic workflows, developers need a way to instrument their applications from the start.

[Datadog Agentic Onboarding](https://docs.datadoghq.com/agentic_onboarding/setup.md?tab=interactivesetup) enables developers and their agents to set up Datadog from the environments where they already work. You can let an agent handle setup, work through the process yourself from a terminal, or ask a coding assistant to instrument your project. Agentic Onboarding detects the technologies in the repository that you want to instrument, prepares the necessary changes, and provides a path to verify that data reaches Datadog.

In this post, we’ll show how Agentic Onboarding helps you:

- [Delegate Datadog setup to your AI agent or guide it yourself](#delegate-datadog-setup-to-your-ai-agent-or-guide-it-yourself)
- [Set up Datadog across applications and infrastructure](#set-up-datadog-across-applications-and-infrastructure)
- [Move from code changes to live telemetry](#move-from-code-changes-to-live-telemetry)

## Delegate Datadog setup to your AI agent or guide it yourself

Agentic Onboarding includes the [Pup CLI](https://docs.datadoghq.com/agentic_onboarding.md?tab=interactivesetup#pup-cli), an agent-ready tool for setting up Datadog from a terminal. You can initiate setup in three ways: run the Pup CLI in headless mode, follow the CLI’s prompts in interactive mode, or ask a coding assistant to instrument your project through the Datadog MCP Server. Each option analyzes your project and prepares the necessary instrumentation changes locally. To illustrate these workflows with an example, we’ll use a Next.js application that we want to instrument with [Datadog Real User Monitoring (RUM)](https://www.datadoghq.com/product/real-user-monitoring/).

### Run setup unattended in headless mode

Headless mode enables an AI agent, script, or CI job to run the Pup CLI directly from a repository without you having to respond to prompts or make interactive selections. For our example application, an agent runs the CLI from the Next.js repository, and the CLI completes the RUM setup unattended. The terminal output reports progress and provides a link to the resulting RUM application in Datadog.

![Output from Pup CLI headless mode showing environment validation and RUM instrumentation progress.](https://web-assets.dd-static.net/42588/1795111854-agentic-onboarding-headless-mode-setup-progress.png)

Fully unattended execution requires the appropriate authentication and authorization. See the[ headless mode documentation](https://docs.datadoghq.com/agentic_onboarding/setup.md#headless-mode) for details.

### Run setup in interactive mode

Interactive mode provides a guided workflow if you want to make setup selections yourself. The Pup CLI runs independently of your own coding assistant or AI model, so setup through the CLI does not consume your model tokens. After you [invoke the Pup CLI](https://docs.datadoghq.com/agentic_onboarding/setup.md?tab=interactivesetup#run-the-setup-command), you can create a Datadog account or authenticate to an existing one. The CLI then asks you to confirm the repository that you want to instrument.

The CLI scans the repository to identify its technologies and checks for uncommitted changes before modifying the project.

![Screen from Pup CLI interactive mode showing the repository scan detecting Node.js, Next.js, and React.](https://web-assets.dd-static.net/42588/1795111922-agentic-onboarding-interactive-mode-scanning-repo.png)

The CLI then recommends supported Datadog products based on the detected stack. You can review those recommendations, change the selection, or explore the other available setup options before asking the CLI to instrument the application.

![Screen from Pup CLI interactive mode recommending RUM and Error Tracking based on the detected stack.](https://web-assets.dd-static.net/42588/1795111967-agentic-onboarding-interactive-mode-recommendations.png)

### Let your coding assistant handle setup with the Datadog MCP Server

The[ Datadog MCP Server](https://docs.datadoghq.com/agentic_onboarding/setup.md#mcp-server) makes onboarding tools available to MCP-compatible coding assistants such as Claude Code, Codex, and Cursor. After you connect your coding assistant to the MCP Server and authenticate with your Datadog account, you can initiate setup by describing what you want. To initiate our example RUM setup through a coding assistant, you can open the Next.js project in your coding assistant and enter the following:

`Add Datadog Real User Monitoring to my project.`

The assistant examines the project, calls the appropriate onboarding tools with your permission, and prepares the instrumentation changes locally.

![Output from an MCP-connected coding assistant reporting that RUM was added to the Next.js application and providing instructions for generating and verifying telemetry.](https://web-assets.dd-static.net/42588/1795112029-agentic-onboarding-mcp-rum-setup.png)

## Set up Datadog across applications and infrastructure

Regardless of how you initiate setup, Agentic Onboarding detects the technologies in your project and prepares code and configuration changes that fit the project’s existing structure. For our sample Next.js application, Agentic Onboarding identifies the application’s Node.js, Next.js, and React components. With RUM as the setup target, it creates the corresponding RUM application in Datadog, adds the [Browser RUM SDK](https://docs.datadoghq.com/real_user_monitoring/application_monitoring/browser/setup/client.md?tab=npm) to the project, and configures the application to begin collecting browser telemetry.

The same project-aware approach extends across supported application frameworks and infrastructure configurations. Agentic Onboarding can instrument application code and work with infrastructure-as-code (IaC) technologies such as Terraform, Helm, Kustomize, and Docker Compose. For specific lists, see the [supported products](https://docs.datadoghq.com/agentic_onboarding/setup.md?tab=interactivesetup#supported-products) and [supported frameworks](https://docs.datadoghq.com/agentic_onboarding/setup.md?tab=interactivesetup#supported-frameworks).

Agentic Onboarding also supports serverless applications. To learn how it adapts instrumentation for AWS Lambda, Google Cloud Run, and Azure Container Apps, see our blog post about[ instrumenting serverless applications with Agentic Onboarding](https://www.datadoghq.com/blog/serverless-agentic-onboarding.md).

## Move from code changes to live telemetry

Agentic Onboarding makes code and configuration changes for your selected Datadog setup directly in your local repository. Your team can inspect the changes, make any necessary adjustments, and move them through its usual testing, approval, and deployment process.

For the Next.js application in our example, the next step is to run or deploy the instrumented application and load it in a browser to generate telemetry. Agentic Onboarding provides a link to the corresponding RUM application, where you can confirm that browser telemetry is arriving in Datadog. From there, you can begin examining frontend performance and user experience.

![Datadog RUM overview showing two page views and frontend performance data from the newly instrumented Next.js application.](https://web-assets.dd-static.net/42588/1795112100-agentic-onboarding-rum-page-view.png)

## Get started with Agentic Onboarding

Agentic Onboarding brings Datadog setup into the workflows where developers and AI agents build software. It helps reduce the manual work required to instrument applications and infrastructure while keeping setup changes visible and reviewable. To get started, [install the Pup CLI](https://docs.datadoghq.com/cli.md#installation) by using `npm install -g @datadog/pup` for the terminal workflow or [set up the Datadog MCP Server](https://docs.datadoghq.com/mcp_server/setup.md) for your coding assistant. For more information, [read the Agentic Onboarding documentation](https://docs.datadoghq.com/agentic_onboarding.md).

If you're new to Datadog, you can <!-- Sign-up trigger (sign up for a 14-day free trial) omitted -->.