​

answers

orc answers lists and retrieves persisted AI answers. Inspect answers by prompt, topic, monitored brand, measurement channel, execution context, time range, and sentiment. The export operation records a CSV generated by the client.

The target is answers in the selected Workspace. To measure new answers, use orc runs. answer is a compatibility alias for the same operations.

List persisted answers.

--brand-id selects the monitored brand owning the prompt, not any brand mentioned in an answer. --platform exactly matches a stored channel ID. Persona is an execution-time value rather than a Persona resource ID lookup; region uses execution, prompt, or snapshot locale, not Workspace residency.

--tag matches active Workspace tag names case-insensitively; supply a name, not an ID. Stored execution tags take precedence, otherwise current prompt tags are used. --brand-mentioned filters by whether the stored mentions array is nonempty; it can differ from the derived brand_mentioned response. Sentiment uses a stored label or numeric score; unscored answers do not match.

Defaults to descending created_at, with answer ID breaking ties. Prompt text sorting and ascending order are also available. Retain sort, order, and filters while following cursors. --start-date is an inclusive UTC timestamp; --end-date is exclusive.

Return answers for this prompt ID. Omit to include all accessible prompts.

Type: string. Optional.

Return answers associated with this topic through the current prompt or its saved observation snapshot.

Type: string. Optional.

Return answers whose topic belongs to this monitored brand. This selects the target brand, not a brand mentioned anywhere in the answer.

Type: string. Optional.

Exact stored observation-channel filter, such as chatgpt-ui. Omit to include all channels.

Type: string. Optional.

Exact persona value stored in the original execution request. This is not a Persona resource ID lookup.

Type: string. Optional.

Observation country or region, matched without case sensitivity. The server uses the execution request, then prompt or saved snapshot locale; this is not workspace residency.

Type: string. Optional.

Tag name, matched without case sensitivity among active workspace tags. Stored request tag IDs take precedence; otherwise current prompt tags are used. Supply a name, not an ID.

Type: string. Optional.

Exact free-form prompt_type stored in the execution request or saved observation snapshot. Omit to include every type.

Type: string. Optional.

Filter by whether the stored answer mentions array is nonempty (true) or empty/missing (false). This filter is based on stored mentions and can differ from the derived brand_mentioned response field.

Type: string. Optional.

Filter by the stored analysis label. If a label is absent, positive and negative numeric scores map to their respective labels and zero maps to neutral. Unscored answers do not match.; enum: positive|neutral|negative

Type: string. Optional.

Inclusive lower bound on created_at as a UTC date-time.

Type: string. Optional.

Exclusive upper bound on created_at as a UTC date-time.

Type: string. Optional.

Sort by creation time (default) or prompt text. The answer ID breaks ties. Keep the same sort and filters while following a cursor.; enum: prompt|created_at

Type: string. Optional.

Sort direction for the selected field and ID tie-breaker. Defaults to desc. Keep this value unchanged while paging.; enum: asc|desc

Type: string. Optional.

Record a client-generated file export using --export-format csv and --row-count. The server does not generate the answer CSV or verify its rows or actual row count.

(required) Client-generated export format. Only csv is accepted.; enum: csv

Type: string. Optional.

(required) Number of rows in the client-generated file. Validated as request metadata; the server does not generate or verify the exported rows.

Type: number. Optional.

Retrieve a persisted answer using the answer row ID (PromptAnswerData.id) returned by listing. Distinguish it from a run ID.

Filter by one channel and prompt.

Retrieve answer details.

Record an already generated CSV export.

Authenticate and select the target Workspace. Access to that Workspace is required.

The following global options can be used with orc answers:

For details and examples, see global options.