> ## 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.

# builtin__decide

> Asks a decision model typed questions about one piece of data.

## Behavior

* Each question is answered independently, in one call.
* Question types: `likelihood` (a probability), `choice` (one option) and `score` (a level on a scale).
* The result maps each question name to its answer, such as `{{E1.urgent.probability}}`.
* Needs a decision model role, `decider` by default.

## When to use it

* Put the data in `state` and keep the questions short.

| Loaded | Kind | Effect |
| - | - | - |
| Always | `decision` | Not marked read-only |

## Input

**state** `any` *required*

The data to judge: text, an object, or a list, usually a template like \{\{E2.ticket}} or \{\{item}}.

**questions** `object` *required*

Question name → question. Likelihood: \{"type": "likelihood", "instructions": "Is it urgent?"}, optionally with "criteria": \{"true": …, "false": …}. Choice: \{"type": "choice", "instructions": "Which team?", "criteria": \{"billing": "Payments", "technical": "Bugs"}}. Score: \{"type": "score", "instructions": "How frustrated?", "criteria": \["Calm", "Frustrated", "Angry"]} with 2 to 10 levels.

**questions.\<name>.type** `enum\<string\>` *required*

Available options: `likelihood`, `choice`, `score`

**questions.\<name>.instructions** `string` *required*

**questions.\<name>.criteria** `any`

```yaml Plan step theme={null}
- id: E1
  tool_name: builtin__decide
  input:
    state: "{{input.state}}"
    questions: "{{input.questions}}"
```

```bash Inspect theme={null}
graph tool show builtin__decide
```


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