v1

latestOpenAPI 3.1.02026-07-2295105298.6 KB

Basics

Start Response Streaming

Kicks off a response-streaming run for a website: Athena queries the selected AI models for the website's prompts and ingests the responses. Omit prompt_ids to run every active prompt, or pass a selection (up to 1,000). Returns immediately with the workflow_id of the run and status: "running"; poll GET /api/v1/response-streaming/status for progress. Requires a website-admin API key.

post/api/v1/response-streaming/start

Request body

website_idstring uuid required

Website to run the response streaming for.

prompt_idsstring[]

Specific prompts to run (1–1,000). Omit to run all active prompts for the website. Only active prompts are executed: paused prompt IDs are accepted (the run still returns 201) but are silently skipped and never queued, so pass active prompt IDs.

selected_modelsobject

Map of model id to enabled flag (e.g. { "ai_overview": true, "ai_mode": false }). Omit to run every currently active model — this is not the dashboard's saved default selection. An explicit selection that resolves to no available models is rejected with 400.

prompt_variationsinteger

Number of paraphrased variations to generate per prompt (0–10).

use_personasboolean

Whether to run the prompts through configured personas.

use_locationsboolean

Whether to run the prompts across the website's configured locations.

only_locationsboolean

When true, skip the base (non-localized) run and execute only the location-scoped variants. Requires use_locations: true: sending only_locations: true without use_locations: true queues nothing (the base run is skipped and no location variants are added), so the run completes as a no-op.

Response

Run started

workflow_idstring required

status'running' required