n8n-nodes-typesafe

# n8n-nodes-typesafe

[n8n](https://n8n.io) community node for [TypeSafe](https://docs.typesafe.ai)
System One models. Ask **typed questions** about your data and get back calibrated
probabilities your workflow can branch on — not free text you have to parse.

## Node: TypeSafe

Operation `Ask Questions` sends a `state` plus a map of questions to
`POST /v1/systemone` and returns one answer per question.

All questions in the node go out in **a single request** and are evaluated in
parallel against the same state. That is the point: per TypeSafe's
[parallel questions cookbook](https://docs.typesafe.ai/cookbooks/parallel_questions.md),
batching is dramatically cheaper and faster than one call per question, so put
independent questions in one node rather than chaining several.

### Question types

| Type | Answer | Use for |
| --- | --- | --- |
| **Noul** | `noul`: probability 0–1 that the answer is yes | whether a condition holds |
| **Choice** | `choice` + `probabilities` per option + `confidence` | picking one option from a set |
| **Score** | `score` (probability-weighted, can land between levels) + `legend` + `probabilities` + `confidence` | rating against ordered levels |

Options are written one per line as `option = description` (the description is
optional). Score levels are one per line, lowest to highest, at least two.

For structured `instructions` or `criteria` — see
[Advanced: structure](https://docs.typesafe.ai/primitives/advanced.md) — switch a
question to **Define as JSON** and supply the whole question object.

### Output

By default each item carries the full response:

```json
{
  "model": "jev-1.13.0",
  "answers": {
    "is_urgent": { "type": "noul", "noul": 0.92 },
    "department": {
      "type": "choice",
      "choice": "technical",
      "probab