Skip to main content
POST
Accept prompt suggestions
Creates prompts from up to 20 suggestions of the latest generation. Accepted prompts are monitored by default, which adds a recurring credit consumption reported in dailyCredits. Set dryRun to true to preview the outcomes and that consumption without creating anything. Requires an active Business, Agency Business or Enterprise plan, and an API token with write access enabled by an Owner or Manager. See Writing data for activation, idempotency, conditional requests and error handling.
string
required
Bearer token for your team.
string
required
Unique key for this operation, at most 255 characters. Completed responses replay for 24 hours.
string
required
Workspace identifier. The workspace must belong to the token’s team.
string[]
required
Between 1 and 20 suggestion identifiers from Get prompt suggestions. Each identifier has at most 200 characters. IDs are opaque and bind to content and taxonomy. Read them again after regeneration. A replaced suggestion returns not_found instead of accepting different content.
string | null
Topic for every accepted prompt. Defaults to the suggested topic.
string[]
At most 20 tag identifiers applied to every accepted prompt.
boolean
default:"true"
Whether accepted prompts take part in scheduled monitoring.
boolean
default:"false"
Validate the request and return the outcomes without creating prompts.

Response

Returns 200 with dryRun, dailyCredits and items. Inspect every item: the HTTP status alone does not mean every suggestion was accepted. already_tracked means that an equivalent prompt is already active and monitored, or that an earlier suggestion of the same request has the same text. A suggestion equivalent to an archived or paused prompt reactivates that prompt and reports created, as in the dashboard Studio. The import rewrites its text, funnel stage and topic, clears its countries, and resets its personas and models to the workspace defaults. Its frequency, boost and last collection are kept. Accepting starts no collection itself. On a LIVE workspace, it can make monitored prompts due at the next scheduler pass, which runs on the hour and on the half hour (UTC): a new prompt when automatic answers are on, or a reactivated prompt whose last collection is older than its frequency. That first collection uses credits once, on top of dailyCredits. This response, dryRun included, does not report that amount. The preview of the accept_prompt_suggestions MCP tool does.

Synthetic request and response

Errors

See Error codes for authentication, plan, idempotency and server errors. A token without write access returns 403 API_KEY_READ_ONLY. On 429, honor Retry-After. If the write limiter is unavailable or writes are suspended, the request returns 503 SERVICE_UNAVAILABLE and nothing changes.