Answer object - Chainshift

These are the field rules for the Answer object returned by the answer retrieval API. Check the schema itself on each endpoint page; this page explains when fields are empty and why.

Fields

Field	Description
`id`	Answer ID (22 characters)
`runID`	ID of the run request that produced the answer. May be absent (`null`)
`promptID` / `query`	ID and original text of the prompt that was run
`model` / `region`	AI model and region the answer was collected from
`location`	Location (a city within the country) the answer was collected from. `null` if the run did not specify a location
`status`	`PENDING` → `SUCCESS` or `FAILED`
`content`	Answer body text
`sources`	Sources the AI cited (URL, title, domain)
`fanouts`	Search queries the AI issued internally
`markdownURL`	URL of the raw answer markdown file
`createdAt`	Creation time (RFC3339)

Empty-value rules

Empty values are often normal behavior rather than a problem. Check in this order.

fanouts is an empty array: fan-out is collected only from ChatGPT, Perplexity, and Claude. Other models always return []. See Supported models.
sources is an empty array: the model did not provide sources for that answer. This is model behavior, and Chainshift returns the collected result as is.
content or markdownURL is empty: check status. There is no body while the answer is PENDING (collecting) or if it is FAILED (collection failed).

FAQ

The same prompt gives a different answer every time. AI answers are non-deterministic. Even the same prompt × model combination can produce different content and sources on each run. That is why repeatCount exists. I want to see the sentences actually quoted from a source. Use include=quotes on Cited URL ranking.