Streamline your next major version upgrade: Introducing AI-powered upgrade skills for Data Center Apps

Marketplace General Discussion
, , , , , ,
1

Hi Marketplace partners & app developers

As we prepare for the next major versions across our Data Center products, we want to ensure you have the tools and support needed to update your apps quickly, efficiently, and with minimal friction.

To help you navigate API updates and underlying platform changes, our team is developing dedicated AI upgrade skills designed to scan your codebases, pinpoint required breaking changes, and guide—or automate—the update process.

Here is a preview of what we’re building, how it works, and how you can get ready today!

:light_bulb: What is the AI upgrade skill for shared platform changes?

The AI upgrade skill is a CLI/repository-level tool that runs locally against your app’s source code. It combines automated codebase analysis with step-by-step AI transformation skills to reduce the time and engineering effort required to prepare for upcoming shared platform changes.

Key capabilities include:

  1. Pre-flight codebase scanning

    • Runs a pre-flight scan that highlights the areas of your codebase likely affected by shared-platform changes and maps each finding to the skill that can address it.

    • Produces a human-readable report you can review before starting any migration.

    • Each skill then confirms the details and applies the changes, so the depth of analysis scales with the migration.

  2. Automated & guided transformations

    • Covers major platform-wide changes, including:

      • Jackson 2 → Jackson 3 migrations

      • Java 25 readiness & Apache HTTP Client updates

      • Core platform dependency upgrades across shared libraries

    • Suggests or directly applies precise code modifications, additions, or deprecation fixes in your codebase.

  3. Human-readable upgrade guidance

    • Every upgrade skill is a readable, step-by-step playbook

    • The skills are written in plain language, so developers can also follow them directly by hand as a migration guide.

image (1)
image (1)1660×1365 268 KB

:spiral_calendar: Expected timeline & availability

  • We will start with platform-wide upgrades, targeting October delivery.

:pushpin: What you can do right now:
Early Access Programs (EAPs)—such as current Confluence & Jira DC EAPs—already include the latest shared platform changes. You can start testing your apps against current EAPs today, and leverage our upcoming AI Skills as soon as Phase 1 drops to accelerate your remaining refactoring.

:speech_balloon: We’d love your feedback!

Our goal is to make readiness for the next major versions as seamless and cost-effective as possible for all Marketplace Partners.
For more context, see our previous announcement: From stable to unshakeable: announcing upcoming major releases for Atlassian Data Center.

  • Are there specific libraries or refactoring pain points you’d like our AI skills to prioritize?

  • Do you have questions about running these skills in your local CI/CD or development environments?

Drop your thoughts and questions in the comments below!
Stay tuned for the public repository link and setup instructions coming soon.

2

Hey @AnnaScisowska,

thanks for sharing that, we’re looking forward to it!

Could you let us know, if this skill is something, which will work with several AI tools, or is it specifically build for Rovo Dev?

3

Hi @m.frank ,

Our goal is to build an agent-agnostic tool rather than tie it exclusively to Rovo Dev. We want developers to be able to leverage these upgrade skills across whichever AI coding assistant or environment they prefer, so we are designing the skills and rule sets so they can be easily integrated into different developer toolchains.

Cheers!

4

Sounds interesting! Since everyone is probably already using Claude Code, and it’s been very efficient for the last releases, I have 3 small points to address, from easiest to hardest:

Who has the ownership of the resulting code?

Who is responsible if the results aren’t fit for production?

Is it possible to release the documentation and examples that you’ve used to create the skill, so that we don’t end up in a situation where the information is only accessible through the skill?

Thank you

5

Hi @aragot.

Great questions — these are exactly the right things to think through, and happy to walk through each.

Who has the ownership of the resulting code?

This is effectively the same as “who owns AI-generated code.” Ownership follows the terms of the model provider that runs the skill.

Who is responsible if the results aren’t fit for production?

The users of the upgrade skills (vendors/developers) are responsible for testing the generated code and confirming it works as expected. Our migration skills are designed to assist with the heavy lifting of upgrade work, but they don’t replace the developer’s engineering judgment — the end results still require manual review, and the decision on whether a change is fit for production is always the developer’s to make. Think of the skills as an experienced pair of hands that accelerates the migration; the final sign-off stays with the developer and their team. We’re always happy to help developers get the most out of them along the way.

Is it possible to release the documentation and examples that you’ve used to create the skill, so that we don’t end up in a situation where the information is only accessible through the skill?

Yes — we’ll add documentation and the relevant public references behind each skill, so the information isn’t locked inside the skill. Some inputs are internal Atlassian material (source code and RFCs) that we can’t republish as-is, but the outcomes map to public developer docs and upgrade guides, and we’ll link those along with worked examples.

Cheers!

6

Hey Anna thanks for sharing! Is there an expected timeline for this? would it be expected with the general access release?

7

hi Tiffany, Beta opens this month - October. GA is expected to follow in November.