We're evaluating a future iOS app that allows users to create and use small, private personal tools inside a single installed app.
A generative model produces declarative UI and data-processing descriptions. The installed app interprets these using built-in functionality. For example:
UI may use HTML/CSS/SVG with JavaScript disabled.
Data calculations may use restricted, read-only SQL over app-controlled local data.
Generated content cannot download or execute arbitrary code or independently access device APIs.
Persistent changes go through predefined actions controlled by the installed app.
These tools are intended for personal use and are not publicly distributed or submitted as separate apps.
We're trying to understand the applicable guidelines before committing to an iOS implementation.
Questions:
-
Under Guideline 2.5.2, would these generated descriptions generally be considered data/configuration interpreted by the app, or could they be considered code that changes app functionality?
-
Would private, per-user tools of this kind fall under Guideline 4.7, even if they contain no JavaScript and are not distributed to other users?
-
If 4.7 applies, how would the software index and universal-link requirements in 4.7.4 apply to private, locally stored tools?
-
Under 4.7.2, would invoking a predefined, user-authorized native action through the host app be considered exposing native platform APIs to the generated tool?
-
Is the single-binary hosting approach discussed in Guideline 4.2.6 relevant to this architecture?
We're not seeking pre-approval, only clarification on which guidelines and boundaries we should design for.
Thank you.