[DOCS-14082] Restructure Agentic Onboarding setup page

content/en/agentic_onboarding/_index.md

@@ -4,5 +4,5 @@ private: true
 ---
 
 {{< whatsnext desc="Explore Agentic Onboarding" >}}
-    {{< nextlink href="/agentic_onboarding/setup/" >}}<u>Setup</u> - Set up the Datadog MCP server to instrument your frontend applications with coding agents.{{< /nextlink >}}
+    {{< nextlink href="/agentic_onboarding/setup/" >}}<u>Setup</u> - Use the AI Setup CLI, the Datadog MCP Server, or task-scoped skills to instrument your applications with Datadog.{{< /nextlink >}}
 {{< /whatsnext >}}

content/en/agentic_onboarding/setup.md

@@ -1,6 +1,6 @@
 ---
 title: Agentic Onboarding Setup
-description: Set up the Datadog MCP server to instrument your applications with coding agents like Cursor or Claude Code.
+description: Set up Datadog instrumentation with the AI Setup CLI, the Datadog MCP Server, or task-scoped skills for Claude.
 
 ---
 
@@ -10,145 +10,175 @@ Agentic Onboarding is in Preview.
 
 ## Overview
 
-Agentic Onboarding lets LLM coding agents instrument your frontend applications for [Error Tracking][3], [Real User Monitoring (RUM)][4], [Product Analytics][5], [Infrastructure Monitoring][8], and [Serverless Monitoring][7] with a single prompt.
+Agentic Onboarding lets LLM coding agents instrument your applications and infrastructure for [Error Tracking][1], [Real User Monitoring (RUM)][2], [Product Analytics][3], [Infrastructure Monitoring][4], and [Serverless Monitoring][5].
 
-Your coding assistant, such as [Cursor][1] or [Claude Code][2], detects your project's frameworks, adds configuration, and provisions required tokens and apps directly from your IDE.
+There are three ways to get started. Pick the one that matches how you work:
 
-## Supported frameworks
-Agentic Onboarding is available for the following frameworks:
-- **Error Tracking, RUM, and Product Analytics**: Android, Angular, iOS, Next.js, React, Svelte, Vanilla JS, and Vue.
-- **Infrastructure Monitoring with Kubernetes**: Terraform, Ansible, Kustomize, and more.
-- **Serverless Monitoring for AWS Lambda**: Terraform, AWS CDK, Serverless Framework, and more.
+| Path | Use when |
+|------|----------|
+| [Setup CLI](#setup-cli) | You want to set up Datadog from a terminal, without an AI coding assistant. |
+| [MCP Server](#mcp-server) | You use an LLM coding assistant (such as Claude Code or Cursor) and want it to handle framework detection and configuration from your IDE. |
+| [Skills](#skills) | You want a focused, single-task workflow in Claude rather than a full MCP toolset. |
 
-## Setup
+The three paths are complementary and work against the same Datadog account. You can install the Datadog MCP Server in your IDE, run the CLI in a terminal, and invoke skills from Claude.
 
-### Install the Datadog Onboarding MCP server
+## Setup CLI
 
-To install the Datadog Onboarding Model Context Protocol (MCP) server, follow the steps for your coding assistant:
+The Datadog AI Setup CLI is a standalone tool that runs from your terminal. Use it when you don't want to install an MCP server, or when you need to do something the MCP flow doesn't cover, such as account bootstrapping.
 
-{{< tabs >}}
-{{% tab "Claude Code" %}}
-{{< site-region region="gov,gov2" >}}
-<div class="alert alert-danger">Agentic Onboarding is not available in the selected site ({{< region-param key="dd_site_name" >}}).</div>
-{{< /site-region >}}
+The CLI can:
+
+- Create a Datadog account end-to-end from the terminal
+- Link an existing Datadog account to your local environment
+- Instrument local infrastructure-as-code (Terraform, Helm, Kustomize, Ansible, Pulumi, raw Kubernetes manifests) by editing files in place
+- Instrument local application code by adding SDK initialization and configuration for supported frontends and backends
+
+### Install and run
+
+1. Run the CLI with `npx`:
+
+    **No Datadog account yet, or want to choose your product interactively**: Run with no arguments to step through account setup and product choice.
+
+    ```shell
+    npx @datadog/ai-setup-cli
+    ```
+
+    **Existing account, installing a specific product**: Pass `--product` to skip product selection.
+
+    ```shell
+    npx @datadog/ai-setup-cli --product <PRODUCT>
+    ```
+
+    Replace `<PRODUCT>` with one of `error-tracking`, `rum`, `product-analytics`, `kubernetes`, `docker`, or `serverless`.
+
+1. Complete the OAuth flow in your browser when prompted. After authentication, the CLI detects your project's frameworks, applies the required configuration, and provisions any necessary tokens.
+
+1. Commit the changes to your repository and set any new environment variables (API keys, application IDs) in your production environment.
+
+1. See [Next steps](#next-steps) to confirm data is flowing.
+
+## MCP Server
+
+The Datadog MCP Server exposes the `onboarding` toolset to any MCP-compatible coding assistant. After you install and authenticate the server, you instrument a project by typing a one-line prompt. The agent reads your code, calls MCP tools (with your permission), applies changes, and verifies the result.
 
-{{< site-region region="us,us3,us5,eu,ap1,ap2" >}}
-1. Open an active Claude Code session with the /mcp command:
+### Supported frameworks
 
-   <pre>
-   <code>claude mcp add --transport http datadog-onboarding-{{< region-param key=dd_datacenter_lowercase >}} "https://mcp.{{< region-param key=dd_site >}}/api/unstable/mcp-server/mcp?toolsets=onboarding"</code>
-   </pre>
+| Product | Frameworks |
+|---------|------------|
+| Error Tracking, RUM, Product Analytics | Android, Angular, iOS, Next.js, React, Svelte, Vanilla JS, Vue |
+| Infrastructure Monitoring (Kubernetes) | Terraform, Ansible, Kustomize, Helm, Pulumi, raw manifests |
+| Serverless Monitoring (AWS Lambda) | Terraform, AWS CDK, Serverless Framework, SAM |
 
-2. Select the MCP server installed in Step 1. You should see a `disconnected - Enter to login` message. Press <kbd>Enter</kbd>.
-3. When you see the option to authenticate, press <kbd>Enter</kbd>. This brings you to the OAuth screen.
-4. After authentication, choose {{< ui >}}Open{{< /ui >}} to continue and grant access to your Datadog account.
-5. Confirm that MCP tools appear under the **datadog-onboarding-{{< region-param key=dd_datacenter_lowercase >}}** server.
-{{< /site-region >}}
+### Install the MCP Server
 
+{{< tabs >}}
+{{% tab "Claude Code" %}}
+In an active Claude Code session, run:
+
+   <pre><code>claude mcp add --transport http datadog-onboarding-{{< region-param key="dd_datacenter_lowercase" >}} "{{< region-param key="mcp_server_endpoint" >}}?toolsets=onboarding"</code></pre>
 {{% /tab %}}
 
 {{% tab "Cursor" %}}
-{{< site-region region="gov,gov2" >}}
-<div class="alert alert-danger">Agentic Onboarding is not available in the selected site ({{< region-param key="dd_site_name" >}}).</div>
-{{< /site-region >}}
+Click the install deeplink for your [Datadog site](/getting_started/site/), then click {{< ui >}}Install{{< /ui >}} in Cursor for the **datadog-onboarding-{{< region-param key="dd_datacenter_lowercase" >}}** server:
+
+<pre><code>{{< region-param key="cursor_mcp_install_deeplink" >}}</code></pre>
 
-{{< site-region region="us,us3,us5,eu,ap1,ap2" >}}
-1. Copy and paste the following deeplink into your browser:
+{{% collapse-content title="Manual configuration" level="h4" expanded=false %}}
+Add the server to `~/.cursor/mcp.json`:
 
-   <pre>
-   <code>{{< region-param key=cursor_mcp_install_deeplink >}}</code>
-   </pre>
+<pre><code>{
+  "mcpServers": {
+    "datadog-onboarding-{{< region-param key="dd_datacenter_lowercase" >}}": {
+      "url": "{{< region-param key="mcp_server_endpoint" >}}?toolsets=onboarding"
+    }
+  }
+}</code></pre>
+{{% /collapse-content %}}
+{{% /tab %}}
 
-2. In Cursor, click {{< ui >}}Install{{< /ui >}} for the **datadog-onboarding-{{< region-param key=dd_datacenter_lowercase >}}** server.
-3. If the MCP server shows a {{< ui >}}Needs login{{< /ui >}} or {{< ui >}}Connect{{< /ui >}} link, select it and complete the OAuth flow. When prompted, choose {{< ui >}}Open{{< /ui >}} to continue and grant access to your Datadog account.
-4. After authentication, return to Cursor and confirm that MCP tools appear under the **datadog-onboarding-{{< region-param key=dd_datacenter_lowercase >}}** server.
-{{< /site-region >}}
+{{% tab "Other MCP clients" %}}
+Any MCP client that supports HTTP transport works. Point it at the endpoint for your [Datadog site](/getting_started/site/):
 
+   <pre><code>{{< region-param key="mcp_server_endpoint" >}}?toolsets=onboarding</code></pre>
 {{% /tab %}}
+{{< /tabs >}}
 
-{{% tab "Datadog AI Setup CLI" %}}
-{{< site-region region="gov" >}}
-<div class="alert alert-danger">Agentic Onboarding is not available in the selected site ({{< region-param key="dd_site_name" >}}).</div>
-{{< /site-region >}}
+### Authenticate the MCP Server
 
-{{< site-region region="us,us3,us5,eu,ap1,ap2" >}}
-The Datadog AI Setup CLI configures your project without a coding assistant.
+1. When prompted to authenticate, press <kbd>Enter</kbd>. This opens the Datadog OAuth screen in your browser.
+1. After authentication completes, choose {{< ui >}}Open{{< /ui >}} to return to your IDE and grant the MCP Server access to your Datadog account.
+1. Confirm that MCP tools appear under the **datadog-onboarding-{{< region-param key="dd_datacenter_lowercase" >}}** server.
 
-1. Run the `npx` command, replacing `<PRODUCT>` with the identifier for the product you want to set up:
+### Instrument your project
 
-   | Product | Identifier |
-   |---------|------------|
-   | Error Tracking | `error-tracking` |
-   | Infrastructure Monitoring | `infra-monitoring` |
-   | Product Analytics | `product-analytics` |
-   | Real User Monitoring | `rum` |
-   | Studio | `studio` |
+Send the prompt that matches the product you want to install:
 
-   ```shell
-   npx @datadog/ai-setup-cli --product <PRODUCT>
-   ```
+{{< tabs >}}
+{{% tab "Error Tracking" %}}
+{{< code-block lang="text" >}}Add Datadog Error Tracking to my project{{< /code-block >}}
+{{% /tab %}}
 
-2. A browser window opens for authentication. Complete the OAuth flow and grant access to your Datadog account.
-3. Return to your terminal. The CLI detects your project's frameworks, applies the required configuration, and provisions any necessary tokens.
+{{% tab "Real User Monitoring" %}}
+{{< code-block lang="text" >}}Add Datadog Real User Monitoring to my project{{< /code-block >}}
+{{% /tab %}}
 
-After the CLI completes, skip to [Deploy your app to production](#deploy-your-app-to-production).
-{{< /site-region >}}
+{{% tab "Product Analytics" %}}
+{{< code-block lang="text" >}}Add Datadog Product Analytics to my project{{< /code-block >}}
+{{% /tab %}}
 
+{{% tab "Kubernetes" %}}
+{{< code-block lang="text" >}}Add Datadog for Kubernetes to my project{{< /code-block >}}
 {{% /tab %}}
-{{< /tabs >}}
 
-### Set up your project
+{{% tab "Docker" %}}
+{{< code-block lang="text" >}}Add Datadog for Docker to my project{{< /code-block >}}
+{{% /tab %}}
 
-Your AI coding agent can help configure Datadog for your project. When you provide a setup prompt, the agent:
+{{% tab "Serverless" %}}
+{{< code-block lang="text" >}}Instrument my AWS Lambda functions with Datadog{{< /code-block >}}
+{{% /tab %}}
+{{< /tabs >}}
 
-- Analyzes your project and identifies the framework, language, and bundler
-- Calls the MCP tool and requests permission before running
-- Applies the configuration changes specified by the tool
-- Provides steps to verify that your application is sending telemetry to Datadog
+The agent detects your stack, requests permission before each tool call, applies changes, and prints verification steps.
 
 **Note**: Your coding agent makes changes locally but does not commit them.
 
-To get started:
-1. Choose the product you want to use and paste its setup prompt into your AI agent:
+After the agent completes, commit the changes to your repository and set any new environment variables (API keys, application IDs) in your production environment. Then see [Next steps](#next-steps) to confirm data is flowing.
+
+## Skills
 
-   {{< tabs >}}
-   {{% tab "Error Tracking" %}}
-   {{< code-block lang="text" >}}Add Datadog Error Tracking to my project{{< /code-block >}}
-   {{% /tab %}}
+Skills are task-scoped capabilities for Claude. The Datadog MCP Server exposes a broad onboarding toolset, but a skill is a focused workflow for a single onboarding step. Use a skill when you only need one thing done and don't want to load a full MCP server.
 
-   {{% tab "Real User Monitoring" %}}
-   {{< code-block lang="text" >}}Add Datadog Real User Monitoring to my project{{< /code-block >}}
-   {{% /tab %}}
+<div class="alert alert-info">This section is in progress. Additional onboarding skills are planned, and the content below is subject to change before general availability.</div>
 
-   {{% tab "Product Analytics" %}}
-   {{< code-block lang="text" >}}Add Datadog Product Analytics to my project{{< /code-block >}}
-   {{% /tab %}}
+### Available skills
 
-   {{% tab "Infrastructure Monitoring" %}}
-   **Kubernetes**
-   {{< code-block lang="text" >}}Add Datadog for Kubernetes to my project{{< /code-block >}}
+| Skill | What it does |
+|-------|--------------|
+| Install Datadog Agent | Detects the host or cluster type and installs the Datadog Agent with the right configuration. |
+| Enable cloud integration | Wires up AWS, GCP, or Azure cloud integrations against your Datadog account. |
 
-   **Docker**
-   {{< code-block lang="text" >}}Add Datadog for Docker to my project{{< /code-block >}}
-   {{% /tab %}}
+### Invoke a skill
 
-   {{% tab "Serverless Monitoring" %}}
-   {{< code-block lang="text" >}}Instrument my AWS Lambda functions with Datadog{{< /code-block >}}
-   {{% /tab %}}
-   {{< /tabs >}}
+Skills are discoverable inside Claude. From a Claude conversation:
 
-2. Review and accept each action your AI agent proposes to complete the setup process.
+1. Reference the skill by name (for example, "install the Datadog Agent on this host").
+1. Claude loads the skill, asks for any required credentials or scope (such as host or cluster type, or cloud account ID), and walks through the steps.
+1. The skill prints verification commands when it finishes.
 
-### Deploy your app to production
+Skills work standalone, so you don't need the MCP Server installed to use them. After the skill completes, see [Next steps](#next-steps).
 
-Commit the changes to your repository and configure the provided environment variables in your production environment.
+## Next steps
 
-[1]: https://cursor.com/
-[2]: https://claude.ai/
-[3]: /error_tracking/frontend/
-[4]: /real_user_monitoring/
-[5]: /product_analytics/
-[6]: /integrations/kubernetes
-[7]: /serverless/
-[8]: /containers/kubernetes/
+- Confirm data is flowing in the Datadog UI: [APM > Services][6], [RUM > Applications][7], [Infrastructure > Hosts][8], or [Logs > Live Tail][9].
+- For team-wide rollout, propagate environment variables through your secret manager.
 
+[1]: /error_tracking/frontend/
+[2]: /real_user_monitoring/
+[3]: /product_analytics/
+[4]: /containers/kubernetes/
+[5]: /serverless/
+[6]: https://app.datadoghq.com/services
+[7]: https://app.datadoghq.com/rum/list
+[8]: https://app.datadoghq.com/infrastructure
+[9]: https://app.datadoghq.com/logs/livetail