I have an Atlassian Forge app that uses Custom UI (React) instead of the UI Kit. Despite updating to the latest Forge CLI and making sure my manifest.yml runtime is set to name: nodejs18, I continue to see this message in Jira:
“App update required
[App Name] version is outdated due to a deprecated platform component. The app developer needs to update it to a compatible version. Please ask your admin to check for updates or contact the app developer.”
I have:
Upgraded to the latest @forge/cli (verified forge --version is 13.x or higher)
Specified runtime.name: nodejs18 in the manifest
Not using UI Kit (I’m using Custom UI)
Confirmed my package.json dependencies are up to date
Yet, the “App update required” banner still appears in Jira. I’m unable to pinpoint which “deprecated component” the warning might refer to. Has anyone experienced this or found a solution? Could there be another deprecated aspect or module I need to update?
Any guidance or troubleshooting tips would be greatly appreciated. Thank you!
Thank you for that - I can’t see anything immediately wrong with your manifest. As you say, no UI kit!
One more thing to check: Make sure that the installation you are testing against is on the latest version of your app (ie. run forge install list and make sure the Version column says Latest not Out of date).
If you are still stuck, the UI team has requested that you create a support ticket @ developer.atlassian.com/support so we can investigate further.
Thank you for your advice. I’ve confirmed that the app is marked as “Latest” using forge install list. However, since the issue persists, I’ve decided to reach out to Atlassian support directly. I plan to attach my manifest.yml and package.json files with my support request.
Experiencing a similar issue, added an app runtime of nodejs22.x, removed @forge/ui from my code. Still, the developer console tells me “Your app has one or more invocation errors caused by usage of a deprecated Forge runtime”.
It turned out that the error message under My apps - invocation metrics triggered on old messages in the metrics. Decreasing the time window excluding the logs before my migration made it disappear. Reading the error message once more also explains this.