Ìfàṣẹ́pọ̀
Fi bọ́ọ̀lù (bearer token) ranṣẹ gẹ́gẹ́ bí kọ́kọ́rọ́ API. Kọ́kọ́rọ́ náà gbọ́dọ̀ ní ìyọ̀ǹda links.edit; a ó kọ̀ ọ́ silẹ pẹlu 403, kii ṣe 404, ti kọ́kọ́rọ́ náà kò bá ní i.
Ojú abánisọ̀rọ̀ yìí kò gba id ajọ kankan. Kọ́kọ́rọ́ rẹ ti mọ ajọ ti o jẹ ti e, a o si fèsì nipa rẹ̀.
Gbiyanju rẹ
Rọ́pọ̀ èyíkéyìí nínú àwọn àmì ìtọ́ka < > pẹ̀lú iye tirẹ̀, àti àmì ìdánimọ̀ bọ́tìnnì náà pẹ̀lú bọ́tìnnì kan láti inú dásibọ̀ọ̀dù rẹ.
curl -X POST https://api.zinndigital.com/v1/ai/recipes/{recipeId}/run \
-H "Authorization: Bearer zdk_live_…"Ṣé o ti wọlé? Iwọ̀n api ní nú ìgbékalẹ̀ rẹ kún id àjọ gidi rẹ ati bọtini tirẹ, o si nṣiṣẹ ibeere na lòdì si api gidi ki o le rii esi gidi na. Ṣí ojú abáná yìí sílẹ̀ nínú kọnsólù API
Àwọn kúlẹ̀kúlẹ̀
**Returns as soon as the run is recorded — it does not wait for the article.** Writing takes longer than an HTTP request may live: the dashboard aborts at 20s and the engine's own worker timeout is 30s, so a synchronous run was killed mid-call every time. The response carries a run whose `state` is `queued`; poll `GET /v1/ai/recipes/{recipeId}/runs` until it reaches `succeeded`, `partial` or `failed`. A queued run normally starts within two minutes. The work then happens on the customer's own key when they have one, on Zinn® AI credits otherwise — **one meter either way** (owner directive D4). What happens after the drafts exist is the recipe's `publish_mode`, which defaults to `draft`: nothing is scheduled or published unless the customer turned that on for that recipe. A `422` means nothing was queued and nothing was charged — funding is checked before the run is recorded, so an unaffordable run is refused here rather than failing later. Requires `links.edit`.
Àwọn ìpìlẹ̀
| Orúkọ | Iru | Tí a nílò | Kini o jẹ |
|---|---|---|---|
recipeId (path) | Uuid | Bẹẹni | — |
Idahun
| Orúkọ | Iru | Tí a nílò | Kini o jẹ |
|---|---|---|---|
id | Uuid | Bẹẹni | UUIDv7 identifier — sortable by creation time (docs/02 §8). |
recipe_id | Uuid | Bẹẹni | UUIDv7 identifier — sortable by creation time (docs/02 §8). |
trigger | string<manual, schedule> | Bẹẹni | — |
state | string<queued, running, succeeded, partial, failed> | Bẹẹni | ⛔ `partial` is a real state, not a rounding of failure: some posts were written and a later step failed. Telling the customer the run failed would make them run it again and pay… |
started_at | object | Bẹẹkọ | — |
finished_at | object | Bẹẹkọ | — |
posts_created | integer | Bẹẹkọ | — |
post_ids | string[] | Bẹẹkọ | — |
anchor_warnings | string[] | Bẹẹkọ | Anchors the model did not place verbatim, aggregated across the run so the customer sees it once instead of opening ten drafts to find out. |
funding | string | Bẹẹkọ | Who paid. `byo` means the article was written on the customer's own key and we charged nothing. |
model | string | Bẹẹkọ | — |
workflow_id | string | Bẹẹkọ | The durable execution, when there is one. |
error_code | string | Bẹẹkọ | A stable machine code for what went wrong, empty on a clean run. ⛔ Set on a `partial` run as well as a `failed` one — `no_target_sites` (the recipe schedules but names no site)… |
error | string | Bẹẹkọ | The engine's own English sentence, for a code a client does not recognise. ⛔ A client SHOULD render its own translated copy from `error_code` and fall back to this only for an u… |
created_at | string | Bẹẹni | — |
Awọn aṣiṣe ti ibudo ipari yii le da pada
401 · 403 · 404 · 422 · 429