Managing Simulink Projects

Managing Simulink Projects

MATLAB projects coordinate Simulink workflows — they manage the MATLAB path (which Simulink uses to resolve model references, data dictionaries, and requirements), track file membership for source control, and provide label-based automation for selecting targets in CI/CD pipelines.

When to Use

• Creating a new MATLAB project for Simulink models
• Adding models, data dictionaries, or requirements to an existing project
• Configuring project paths so Simulink can resolve references
• Setting up labels for automation (test selection, code generation targets)
• Configuring source control and cache folders
• Diagnosing broken model references, missing dictionaries, or failing project health checks

When NOT to Use

• Building or editing model structure (blocks, connections) → use building-simulink-models
• Writing or running tests → use testing-simulink-models
• Drafting requirements content → use generate-requirement-drafts
• Running simulations for analysis → use simulating-simulink-models

Mental Model

Four principles that prevent the most common failures:

1. Registration ≠ path. addFile registers a file with the project for source control tracking. addPath puts a folder on the MATLAB path so Simulink can resolve artifacts by filename. These are independent operations — you almost always need both.

2. Filename-only resolution. Data dictionaries (set_param(...,'DataDictionary',...), addDataSource), model references, and requirements all resolve via MATLAB path using filename only — never full paths. The containing folder must be on the project path.

3. addPath is not recursive. Each subfolder containing resolvable artifacts needs its own addPath call.

4. Single-project constraint. Only one project can be open at a time. Creating or opening a new project silently closes the current one (running its shutdown scripts). currentProject throws an error when no project is open — it does not return empty.

Guardrails

Always:

• Use addPath(proj, folder) for every folder containing dictionaries, referenced models, requirements, or profiles
• Pass filename only (not full paths) to addDataSource, set_param(...,'DataDictionary',...), and slreq.createLink
• Use absolute paths for SimulinkCacheFolder and SimulinkCodeGenFolder: proj.SimulinkCacheFolder = fullfile(proj.RootFolder, 'work')
• Call updateDependencies(proj) before querying proj.Dependencies
• Close the current project before creating or opening another
• Use .graphml extension for DependencyCacheFile
• Use .gitignore for excluding derived files from source control
• Add requirements folder to project path when linking requirements to models
• Register .slmx link-store files with the project (guarded by isfile check)

Ask First:

• Deleting files from a project (removeFile then disk delete)
• Modifying startup/shutdown scripts (affects all users who open the project)
• Changing project references (affects dependent projects)

Never:

• Use proj.IgnoredFilePatterns — throws "Feature not supported" for all project types from createProject
• Use cat.Labels — the correct property is cat.LabelDefinitions
• Use 'string' as a label DataType — valid types are 'none', 'char', 'double', 'integer', 'logical'
• Use findFile (singular) for label queries in R2024a+ — use findFiles (plural). ONLY use findFile with R2023a/b
• Pass full file paths to filename-only APIs
• Attempt to open two projects simultaneously
• Use clear all in startup scripts — it wipes the proj variable

Task Routing

Verification

Run after every project modification:

matlabCopy code
results = runChecks(proj);
updateDependencies(proj);
deps = proj.Dependencies;  % digraph

If runChecks reports issues, inspect which folders are missing from the project path — this is the root cause of most failures.

Copyright 2026 The MathWorks, Inc.