Skip to main content
Stratos AI turns one app into a stream of TikTok-native content — photo slideshows, illustrated decks, AI-creator videos — and posts it on a schedule. The v1 API exposes that same engine to your own code, an MCP-connected agent, or the stratos CLI.

​
What you can do

Import an app

Paste an App Store link, a website, or just a name. Free, no credits spent.

Generate content

Slideshows, illustrated decks, or hook videos — any topic, any archetype.

Schedule to TikTok

Drafts by default. Direct publish is available, scope-gated.

Read analytics

Views, likes, comments — the same numbers the dashboard shows you.

​
The one rule that shapes everything else

Nothing schedules until it’s approved. Every piece of generated content starts as approvalStatus: "pending". Scheduling a post requires calling POST /v1/content/{id}/approve first — the API rejects the schedule request otherwise, with a 409 that says exactly why. This is the actual safety mechanism against an agent looping generation-and-posting unsupervised — not a rate limit, not a spend cap (though those exist too), the approval step itself. Read The approval gate before building anything that generates and schedules content in the same flow.

​
What doesn’t exist here

Being upfront about this matters more than it sounds: an agent (or the person operating it) should never assume a capability exists because it seems like it should. Two honest gaps, as of this API’s first release:
  • Illustrated decks and hook videos aren’t generation-complete yet. POST /v1/content with type: "illustrated" or type: "hook_video" creates a record but returns a clear generation_not_wired notice instead of a result — their pipelines are asynchronous Durable-Object state machines that need a dedicated integration pass. type: "slideshow" is fully live.
  • Posting requires rendered images, and image compositing is a client-side step today. POST /v1/schedules will fail with a 409 (no_images) on content that has no images array in its result — which today means every slideshow, since text generation and image rendering are separate steps and only the former is wired via this API so far.
Check GET /v1/jobs/{id} or the notice/error fields on any response — the API always tells you which of these you’ve hit, rather than returning a silent partial success.