Request Description
Changed onRun the description (Gemini analysis) task on an asset, or on a folder.
Requires WRITE access on the asset or admin.
For a single asset, needs_downscale reports whether a player/Gemini- compatible video rendition is missing. When it is, we encode a compatible proxy first and chain the description task after it (the downscale's link callback), so Gemini only runs once a usable rendition exists. The returned task_id always tracks the description task, so callers poll the same id in both cases.
Pointed at a folder, it queues a walk of that folder instead and returns as soon as the walk is queued: task_id then tracks the fan-out, which dispatches one description task per asset — each exactly as it would be dispatched on its own, proxy encode included — and needs_downscale is false, since which assets need one is only known once the walk runs. The folder is walked recursively; assets the caller cannot write, and assets whose media is not uploaded, are skipped. There is no limit on its size, and every dispatched asset is billed — the preflight route reports how many that will be before you commit to it.
An optional body forces the configuration to analyse under, so a one-off run does not need a tellers-config committed to the folder tree first. It replaces what the tree resolves rather than merging over it. Every entity id in it is checked against the caller before anything is dispatched, and the whole request is refused if one is not theirs. The same body on the preflight route shows the resulting config first.
Request
- Base URLhttps://api.tellers.ai (the document declares no server; this is the origin it was published from)
- Authnone declared
Path parameters
Headers
Request body
Response
Successful Response
Changes
- info
added the new optional request property
////
- info