I hope these suggestions are helpful, and inspire both conversation and action!
The barrier of entry for creating a Nextcloud app has never been lower. Over the last couple years there has been a massive increase in the number of pull requests (and applications) accepted into the appstore, directly related to wider AI coding availability. Not judging it, so much as acknowledging how offering a standards-compliant platform makes such rapid AI development inevitable. There is no reason to assume this process will ever slow down… which is why app developer support scaffolding must scale up accordingly. My question is:
What is being done to manage submissions, scan for vulnerabilities, confirm API compliance, mitigate redundancies, and protect Nextcloud’s appstore reputation?
An example:
I’ve noticed Nextcloud Social and Euro Office receiving extensive pull requests from AI-assisted Nextcloud developers, but the pull requests are accompanied by huge numbers of tests, linting and code validation. I do not believe it is yet possible for casual developers to actually follow the Nextcloud Code of Conduct requirements through the current testing framework, nor do I believe their AI assistants are even aware of guardrails in regards to generated code. A developer with a failing app can currently receive a 2 year ban from the appstore, but at the rate newly coded apps are being pushed… it is Nextcloud itself that will suffer from future vulnerabilites in 3rd party applications within their ecosystem… I also worry about malicious code injection within the greater platform, weaponized within the large codebase.
What can be done to centralize linting standards and decentralize test execution?
Perhaps official boilerplate that mirrors the test structures of Nextcloud Social, and other core apps, could be made available for developers to test locally (explicitly designed for ephemeral AI reference). Or guidelines for developers to improve their personal CI pipelines. Providing clarification to external developers could at least reduce the volume of mistakes before submission: both for native php and external applications. The “Code of Conduct” could require more extensive “App Publishing Guidelines” that developers must follow, designed as local testing guides for local AI recognition, but also as a boilerplate reference for the AI to then explain to the developer why the code they are generating isn’t acceptable for the Nextcloud appstore and ecosystem.