> ## Documentation Index
> Fetch the complete documentation index at: https://graph.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# workbench__set_interview

> Shows the architect's interview in the pane and ends its turn with its question.

## Behavior

* Carries the reply you read, the architect's confidence, what's still missing, the question with its guess, and the facts so far, each known, a guess, being asked or missing.
* Ends the architect's turn, so a question costs one model call.

| Available to | Effect |
| - | - |
| Workbench agents only | Read-only |

## Input

**confidence** `integer` *required*

**confirmed** `boolean`

True once the user said yes to the restate

**facts** `object[]` *required*

**facts.\[].label** `string` *required*

Outcome, Who, Where, Success, Constraint, Out of scope, or a topic of this request

**facts.\[].previous** `string`

When feedback reopens a confirmed fact: what it said before

**facts.\[].state** `enum\<string\>` *required*

Available options: `known`, `guess`, `asking`, `missing`

**facts.\[].text** `string` *required*

**guess** `string`

Your guess at its answer, with the reasoning

**missing** `string`

What you still don't know, in a few words; empty when nothing is

**question** `string`

The one question you're asking now, when you're asking one

**reply** `string` *required*

Your whole reply to the user, as they read it in the chat: your read of the request, the question and your guess, or the restate

## Output

**confidence** `integer`

**shown** `boolean`

```json Output theme={null}
{
  "confidence": 55,
  "shown": true
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.