Deploy to AEM as a Cloud Service with confidence

Last update: June 8, 2026

https://video.tv.adobe.com/v/3480340/?learn=on&enablevpops

Deployment confidence comes from knowing your environment is healthy before you push. This walkthrough shows how to check AEM environment status, review pipeline history, and trigger deployments from an AI client using the AEM Cloud Manager MCP Server, so teams can move fast without losing visibility.

Scenario details

CX Enterprise applications

Adobe Experience Manager Cloud Manager

Agentic tools

AEM Cloud Manager MCP Server

Audience

Developers, DevOps, operations teams

Prerequisite

MCP-compatible AI client, AEM Cloud Manager access

Each step shows one representative prompt and an example AI response. A More prompts to try section follows for additional exploration in the same session.

Before you begin

Claude Code

Navigate to your project directory first, then add the Cloud Manager MCP Server using the CLI:

code language-bash
`claude mcp add --transport http adobe-cloud-manager https://mcp.adobeaemcloud.com/adobe/mcp/cloudmanager`

Or add it manually to .mcp.json in your project root:

code language-json
`{ "mcpServers": { "adobe-cloud-manager": { "type": "http", "url": "https://mcp.adobeaemcloud.com/adobe/mcp/cloudmanager" } } }`

Restart Claude Code. The Cloud Manager tools are available in your next session.

Full setup: Claude Code MCP documentation

Cursor

Add the Cloud Manager MCP Server to ~/.cursor/mcp.json (global) or .cursor/mcp.json in your project root:

code language-json
`{ "mcpServers": { "adobe-cloud-manager": { "type": "http", "url": "https://mcp.adobeaemcloud.com/adobe/mcp/cloudmanager" } } }`

Open Settings > MCP, select Connect next to the server, and sign in with your Adobe ID.

Full setup: Cursor MCP documentation

GitHub Copilot

Add the Cloud Manager MCP Server to .vscode/mcp.json in your project root:

code language-json
`{ "servers": { "adobe-cloud-manager": { "type": "http", "url": "https://mcp.adobeaemcloud.com/adobe/mcp/cloudmanager" } } }`

Note: VS Code uses "servers" as the top-level key, not "mcpServers".

Open the GitHub Copilot Chat panel, switch to Agent mode, and select Connect next to the server. MCP tools are only available in Agent mode.

Full setup: VS Code MCP servers documentation

Other AI clients

Using another MCP-compatible environment? Connect to the Cloud Manager MCP Server using this endpoint:

code language-none
`https://mcp.adobeaemcloud.com/adobe/mcp/cloudmanager`

Full setup instructions for all supported clients: Connect to your AI client

NOTE

Sign in with your Adobe ID when prompted and select the IMS organization linked to your AEM as a Cloud Service program. Permissions are enforced at the Cloud Manager level. Your AI client can only perform operations your account is authorized for.

On first connection, your AI client may ask you to confirm your organization or AEM program. Once that context is set, the MCP server uses it for the rest of the session.

Some tools prompt for your approval before they execute. Review the proposed action and approve or decline. No action is taken without your confirmation.

Step 1: Check environment status

Before kicking off a release, confirm your environments are healthy and nothing is actively running.

What is the status of the production environment?

See an example response

AI client showing production environment status from Cloud Manager {modal="regular"}

Select to zoom.

Step 2: Review pipeline runs

Review recent pipeline history to understand deployment patterns and catch failures before they block your next release.

Show me the last five pipeline runs for the production pipeline.

See an example response

AI client showing the last five pipeline runs for the production pipeline {modal="regular"}

Select to zoom.

Step 3: Trigger a pipeline

Kick off a pipeline run directly from your AI client. The server confirms the target environment and asks for approval before starting.

Run the Fullstack pipeline against dev environment of WKND sandbox program.

See an example response

AI client showing pipeline trigger confirmation and Cloud Manager UI reflecting the running pipeline {modal="regular"}

Select to zoom.

CAUTION

The AI client will ask you to confirm the pipeline name before triggering a run. Enter the exact pipeline name to proceed. Review the target environment carefully before confirming, especially for pipelines that deploy to production.

Step 4: Check pipeline status

After triggering a run, ask your AI client for a status update without switching to the Cloud Manager interface.

What is the status of the triggered pipeline?

See an example response

AI client showing the status of the triggered pipeline run {modal="regular"}

Select to zoom.

What you accomplished

You used the AEM Cloud Manager MCP Server to check environment health, review pipeline history, trigger a deployment, and verify its status, without opening the Cloud Manager interface. By combining environment visibility and deployment control in a single AI session, development and operations teams can respond to issues faster and keep their workflow inside the tools they already use.

More you can accomplish

The Cloud Manager MCP Server handles far more than what the walkthrough above covers. Expand a scenario below to see prompts you can try in the same session.

Catch problems before a release goes out

Deployments often fail for reasons that were visible before the pipeline ran. These prompts help you confirm environment health, check for conflicting runs, and verify version alignment across environments before you commit to a release.

Prompts

code language-none
`We're about to kick off a production release. Give me a full status check on all environments first.`

code language-none
`Is there anything currently running in the staging pipeline? I don't want to queue on top of an active run.`

code language-none
`Before I promote main branch to production, confirm main was deployed to Dev and all environments are on the same AEM version.`

code language-none
`What repositories are connected to the WKND program?`

Course-correct a deployment that's already in flight

An accidental trigger or a stalled approval gate can cascade into a blocked pipeline or an unwanted deploy. These prompts let you cancel or advance a running pipeline without switching to the Cloud Manager interface.

Prompts

code language-none
`The staging pipeline kicked off by mistake. Cancel it before it deploys.`

code language-none
`The release pipeline is waiting at the approval gate. Advance it to continue the deployment.`

Understand your deployment track record

Knowing when things last succeeded, how long pipelines run, and whether patterns are changing helps you plan releases and catch slow degradation before it becomes an incident. Use these prompts to pull that history on demand.

Prompts

code language-none
`What is the status of the last production pipeline execution? If it failed, explain why.`

code language-none
`When was the last successful deployment to the staging environment?`

code language-none
`Our pipeline times are creeping up. What's the longest run we've had in the last 30 days?`

Get a broken build back on track

When a pipeline fails, the fastest path to resolution is understanding exactly where it broke and why. These prompts surface failure details, change history, and quality gate issues so your team can diagnose and fix without manually digging through logs.

Prompts

code language-none
`We're seeing a regression on the live site. What changed in production over the last week?`

code language-none
`Which pipelines have failed in the last 7 days, and at what stage did they fail?`

code language-none
`The last pipeline failed at the code quality step. What specific issues need to be fixed before I can retry?`

code language-none
`Pull the step logs for the last failed run. I need to see exactly what the quality gate flagged.`

Further information

Resource

What you will find

AEM as a Cloud Service documentation

Full AEM application documentation

recommendation-more-help

cx-enterprise-agentic-tools-help