> For the complete documentation index, see [llms.txt](https://help.modelreef.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.modelreef.io/syntax/how-input-fields-work/auto-validation-rules.md).

# Auto-Validation Rules

This article explains how **auto validation rules** work in Model Reef input fields.

You will learn:

* What types of validation the system performs automatically.
* How bounds, types and patterns contribute to validation.
* How to respond to validation warnings.

Auto validation is designed to catch obvious issues early without blocking legitimate but unusual scenarios.

***

## 1. Types of auto validation

Model Reef applies several kinds of checks when you enter or change values:

* **Type checks**
  * Numeric fields must contain numeric values.
  * Percent fields expect values in an appropriate range.
* **Bounds checks**
  * Values are compared against any configured minimum and maximum bounds.
* **Pattern checks**
  * Large jumps or sign flips compared to recent periods may be highlighted.
  * Values very different from historical actuals may be flagged.

These checks are lightweight and run as you edit.

***

## 2. How validation feedback is shown

When validation detects a potential issue, the interface may:

* Highlight the cell or field in a warning colour.
* Display a tooltip or message explaining the issue.
* Show an icon or marker in the margin of a panel.

Validation messages are advisory; they do not by themselves change data.

***

## 3. Interaction with bounds and units

Auto validation rules use:

* **Bounds**, if defined, as thresholds for warnings.
* **Units and display scale** to ensure that checks are applied in the correct context.

This is why setting realistic bounds and consistent units improves the usefulness of validation feedback.

***

## 4. Structural validation

Beyond numeric values, some structural changes trigger validation, for example:

* Changing variable type from one category to another.
* Moving variables between branches.
* Reclassifying Data Library entries.

In these cases, the system may request confirmation and explain the implications for financial statements.

***

## Dealing with validation warnings

{% stepper %}
{% step %}

### Read the message

Read the message to understand what triggered it.
{% endstep %}

{% step %}

### Check intent

Check whether the value or change is intentional.
{% endstep %}

{% step %}

### Keep if correct

If the value is correct but genuinely outside previous patterns, you can usually keep it.
{% endstep %}

{% step %}

### Correct if mistaken

If the value reflects a mistake, correct it and confirm that the warning disappears.
{% endstep %}
{% endstepper %}

Warnings are there to support judgement, not to replace it.

***

## 6. Validation and collaboration

Validation is especially helpful in collaborative models:

* It reduces the chance that one collaborator introduces extreme or inconsistent values unnoticed.
* It gives Owners and senior Editors more confidence in the integrity of inputs.
* Combined with notes, tags and attachments, it forms part of the model's governance toolkit.

Auto validation plus human review produces the best results.

***

## Related articles

* [Formula Validation Errors](/syntax/formula-syntax/formula-validation-errors.md)
* [Period Toggle Rules](/syntax/chart-and-table-syntax/period-toggle-rules.md)
* [Archive Rules](/help/permissions-and-collaboration/archive-rules.md)
* [Build a Forecast from Ticker Fundamentals](/how-tos/data-workflows-and-automation/build-a-forecast-from-ticker-fundamentals.md)