Dynatrace provides an upgrade skill that helps a coding agent guide you through parts of the upgrade to Latest Dynatrace. The skill isn't a wizard: it doesn't do all the upgrade work for you. You still need to understand the new concepts, decide what the upgrade means for your environment, and complete steps that the skill doesn't cover. For the decisions to make before you start, see Plan your upgrade to Latest Dynatrace.
The upgrade skill provides a coding agent the mapping knowledge it needs to migrate a Dynatrace Classic configuration to Latest Dynatrace. With that knowledge, the agent determines what each Dynatrace Classic object is intended to achieve and proposes a replacement for your review. The following sections explain why this approach suits the upgrade and where you can use the skill.
There's rarely one mechanical mapping from a Dynatrace Classic object to its replacement. Understanding what each configuration was meant to achieve is key to rebuilding it correctly in Latest Dynatrace. A coding agent is good at this kind of work.
Use the upgrade skill to:
The upgrade skill produces a replacement configuration for review and analysis, not changes that have already been applied to your environment. It saves the configuration to files, so you can review it before you deploy anything. The following sections explain how to deploy the configuration and what additional output the skill produces.
The generated configuration uses the raw Settings format, which your existing Configuration as Code tools already accept. You choose how to deploy it:
POST request to the relevant Settings endpoint.In addition to the replacement configuration, the upgrade skill produces assessments, plans, and status tables that document what it found and what it did. Depending on the size of the upgrade, you get the following types of output:
In your environment, where the agent can write to your environment:
Approval prevents the agent from making unwanted changes to your environment. This is especially important because deleted Dynatrace Classic objects can’t be restored through rollback. You can revert an environment's upgrade group during the grace period, but Dynatrace Classic objects that were deleted stay deleted. For details, see Rollback and the grace period.
The upgrade skill runs in your own development environment, not in Dynatrace. Set up the tools it runs in and the access it needs before your first upgrade session.
Meet the following prerequisites:
npx skills add dynatrace/dynatrace-for-ai command.write scopes only when you apply changes.Work in a folder under version control. The skill creates a working branch and commits your changes after each verified task, so that you can undo one task without losing the others. It never pushes changes and never opens a pull request.
You can start an upgrade session in two ways: copy a prompt from the Upgrade Assistant or write your own prompt.
Open the Upgrade Assistant, select the task that you want to perform, copy the prompt from the task, and paste it into your coding agent. For more information, see Get started with the Upgrade Assistant.

Start with one of these types of prompts:
Questions: It is for targeted questions about the upgrade.
Prompt examples:
Single tasks: These are for a scoped task, setup, or an edit.
Prompt examples:
Large upgrade: These are for multiple areas.
Examples:
For details on how upgrade sessions work and when they stop, see Run an upgrade skill session.
The upgrade skill speeds up the upgrade process, but it can't guarantee that a migrated configuration behaves like the Dynatrace Classic one. Review the following limitations, and check the result against your own environment before you rely on it.
The configuration produced by the upgrade skill is generated by an AI coding agent that reasons about your Dynatrace Classic configuration. Review every generated object against your intent before deploying it, and keep the Dynatrace Classic object until you've confirmed that the replacement behaves as expected.
A successful write proves that the payload was well-formed, not that it replicates the Dynatrace Classic behavior.
dtctl apply --dry-run is a client-side preview only.
It reports success without contacting the API, even for a schema that doesn't exist.
It confirms only that the YAML or JSON can be parsed.
When the agent suggests an equivalent based on name matching rather than a documented mapping, it says so. Such a suggestion is based on matching names and keywords, not on verified feature parity, so confirm that it covers what you need.
Detection options, thresholds, and behavior differ between Dynatrace Classic and Latest Dynatrace, so an upgraded object can be correct and still behave differently. For details, see Classic capabilities and where to find them in Latest Dynatrace.
An upgrade can affect your billing costs by changing your consumption.
For example, when you upgrade RUM anomaly detection, custom metric alerts move from non-billable to billable consumption, and baseline changes from percentile(90) to percentile(75) increase your overall platform consumption.
To get reliable results from an AI-assisted upgrade and keep your environment safe, follow these practices:
No. The upgrade guides are standalone.
Almost certainly not. A large share of Dynatrace Classic configuration is no longer used and should be deleted instead.
Yes.
The output uses the Settings format, so that you can deploy it with Monaco, Terraform, dtctl, or a direct POST request.
For details, see Configuration as Code overview.
You review every generated object before it's applied. Nothing is written without your approval, and classic objects aren't deleted before their replacements are verified.
It stops after three failed attempts and escalates. A missing permission is reported as a blocker, with the specific scope named.