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

# assignment

> Set initial variables on the run context before the workflow's substantive work begins.

`assignment` nodes are how you initialise variables. Each assignment has a `name` and an `expression` evaluated once and written to the top-level run context.

## Configuration

```jsonc theme={null}
{
  "id": "asg-1",
  "type": "assignment",
  "data": {
    "label": "Initialize Variables",
    "description": "Set up initial variables for the pipeline",
    "assignments": [
      { "name": "risk_score",         "expression": "0" },
      { "name": "rejection_reasons",  "expression": "[]" },
      { "name": "processing_start",   "expression": "datetime.now()" }
    ]
  }
}
```

## Expression evaluation

Expressions are small Python-like literals or function calls. The supported surface is intentionally narrow:

| Expression                              | Result                                                                                                  |
| --------------------------------------- | ------------------------------------------------------------------------------------------------------- |
| `0`, `'pending'`, `[]`, `true`, `false` | JSON literals                                                                                           |
| `datetime.now()`                        | ISO-8601 datetime string at evaluation time                                                             |
| `uuid()`                                | New v4 UUID                                                                                             |
| `{{some_variable}}`                     | Interpolation from the run context (rarely needed in assignment — use the variable directly downstream) |

For anything more complex, prefer the [`code`](/workflows/nodes/code) node.

## Why initialise

* **Accumulator variables** (`rejection_reasons: []`) that downstream nodes append to via `code` transforms.
* **Tracking timestamps** (`processing_start`) for SLA calculations.
* **Status placeholders** (`kyc_status: 'pending'`) that change as the workflow advances.

A simpler workflow can skip the assignment node entirely and let rules + decision tables write directly to the top-level context — more compact, fewer moving parts.

## Rule

`rule` is the workhorse downstream of `assignment` — single-pass first-match conditions producing named output fields. See [the rule node](/workflows/nodes/rule).