Apple SiriKit Cloud Media

What authentication does SiriKit Cloud Media use?

The SiriKit Cloud Media spec does not declare a standard securityScheme - Apple authenticates the calling Siri service against the vendor-supplied extension URL using a relationship registered in the developer portal. Implementers typically validate the request signature and pin the source IP at the edge.

Who calls the SiriKit Cloud Media endpoints?

Apple's Siri service calls them. The vendor (the music or podcast provider) implements the six endpoints; they do not consume them. The OpenAPI spec is therefore most useful for scaffolding an extension server, generating handler stubs, and writing contract tests against the expected request and response shapes.

What media types does the cloud extension support?

The /configuration endpoint declares supported media types - typically song, album, playlist, station, audiobook, and podcast - and the affinity verbs the vendor honours. Siri uses the response to filter which voice intents it routes to your extension.

How do I scaffold a SiriKit Cloud Media server through Jentic?

Run pip install jentic, search 'sirikit cloud media playMedia', and load the operation to retrieve the request and response schemas. Use those schemas to generate handler stubs in any framework - FastAPI, Express, or Spring - and validate incoming Siri payloads against them.

Are the six endpoints stable across iOS releases?

The shape has been stable since the public SiriKit Cloud Media introduction. Apple iterates on supported media types and affinity verbs through the /configuration response, not by adding new endpoints. Treat the six paths as a fixed contract and feature-flag through configuration.

Can I limit what my agent is allowed to do with the SiriKit Cloud Media API?

Yes. Because you run Jentic One yourself, your own rules decide which of the fixed intent and queue operations your agent may call, so you can allow only what a task needs, such as addMedia or playMedia. Since each operation carries its target in the request body, you can keep affinity or queue-reporting calls like updateMediaAffinity or queues/updateActivity out of scope unless you deliberately add them. Any shared test or staging secret is held encrypted by your own instance and injected at execution time, so it never reaches the agent's prompt or logs.