Model Checker

Model Checker is an Automate function that validates Speckle objects against configurable rules. This approach provides a flexible way to implement quality checks and maintain consistent standards across projects.

Overview

The Model Checker allows you to:

• Define validation rules for your objects
• Configure severity levels for issues
• Check properties across different types of objects
• Generate reports of validation results
• Apply consistent standards across projects

Getting Started

1. Access the Model Checker Application

1. Go to the Model Checker Application
2. Sign in with your Speckle account
3. Create and manage your validation rules through the intuitive web interface

2. Create an Automation

1. Go to your workspace project in Speckle
2. Create a new Automation
3. Select the Model Checker function
4. Configure the function:
   • Set minimum severity level to report
   • Configure other options as needed
5. Save and run your automation

Rule Definition Format

Rules are defined with the following components:

Logic	Property Name	Predicate	Value	Message	Report Severity
WHERE	category	matches	Walls	Wall thickness check	ERROR
CHECK	Width	greater than	200
WHERE	category	matches	Columns	Column height check	WARNING
AND	height	in range	2500,4000

Component Explanation

• Logic: Defines how conditions are combined (WHERE, AND, CHECK)
• Property Name: The object property or parameter to check
• Predicate: Comparison operation (equals, greater than, etc.)
• Value: Reference value for comparison
• Message: Description shown in validation results
• Report Severity: ERROR, WARNING, or INFO

Supported Predicates

Predicate	Description	Example
exists	Checks if a property exists	height exists
equal to	Exact value match	width equal to 300
not equal to	Value doesn't match	material not equal to Concrete
greater than	Value exceeds threshold	height greater than 3000
less than	Value below threshold	thickness less than 50
in range	Value within bounds	elevation in range 0,10000
in list	Value in allowed set	type in list W1,W2,W3
contains	Property contains substring	name contains Beam
does not contain	Property doesn't contain	name does not contain temp
is true	Boolean property is true	is_structural is true
is false	Boolean property is false	is_placeholder is false
is like	Loose text matching	name is like Wall matches Walls

Rule Logic

• WHERE: Filters objects to check (like SELECT WHERE in SQL)
• AND: Additional filter conditions
• CHECK: Final check condition (optional, defaults to last AND)

Objects pass a rule when they match all conditions. Objects that match WHERE/AND filters but fail the CHECK condition are reported as issues.

Working with Object Properties

The Model Checker understands properties in Speckle objects regardless of schema:

• Direct properties: category, name, id
• Nested properties: parameters.WIDTH.value
• Revit parameters: Use parameter names like Mark, Width, Assembly Code

Example Rules

Wall Thickness Check

Rule: WHERE category equals "Walls" AND width less than "200"
Message: "Walls must have width of at least 200."
Severity: ERROR

Door Naming Convention

Rule: WHERE category equals "Doors" AND name is not like "^D\d{3}$"
Message: "All doors must have a name that follows the format "D" followed by three digits."
Severity: WARNING

Structural Column Height Range

Rule: WHERE category equals "Columns" AND is_structural is true AND height not in range "2400,4000"
Message: "Structural columns must have a height between 2400 and 4000."
Severity: ERROR

Support

For issues or questions, please let us know on the Speckle Community Forum.

Alternative: TSV File Format

While the Model Checker Application is the recommended way to create and manage rules, you can also create compatible TSV (Tab-Separated Values) files manually. This can be useful for:

• Programmatically generating rules
• Version controlling rules in a text format
• Integrating with existing workflows
• Creating rules in bulk

The TSV file should follow the same structure as shown in the table above, with columns separated by tabs. The file will then need to be hosted somewhere and served with MIME-type of text/tab-separated-values and the URL used in the automation configuration.