Matlab Run Tests

Run Tests

Run MATLAB test suites, analyze results, collect code coverage metrics, and configure CI/CD pipelines using matlab.unittest infrastructure.

When to Use

• User wants to run an existing test suite
• User needs to filter tests by name, tag, or folder
• User wants to analyze test failures or incomplete results
• User needs code coverage collection or analysis
• User wants CI/CD pipeline configuration with buildtool
• User asks about parallel test execution

When NOT to Use

• Writing tests, creating test classes, parameterization, fixtures, mocking — use matlab-write-tests
• Baseline/regression tests against stored reference data — use matlab-write-tests
• Testing Simulink models — use Simulink test skills
• Performance benchmarking — use profiling workflows
• Buildtool setup without test context — use matlab-software-development:matlab-create-buildfile

Workflow

Run and report

1. Run tests — Execute via run_matlab_test_file MCP tool or runtests with filtering
2. Analyze results — Report pass/fail counts, iterate failed tests for diagnostics
3. Verify — Confirm all tests pass before proceeding to coverage or CI

Coverage analysis (when requested)

1. Collect — Set up TestRunner with CodeCoveragePlugin and CoverageResult
2. Report summary — Call coverageSummary and present percentages to user
3. Act on results — Generate reports, analyze gaps, or justify — as user directs

CI/CD setup

1. Create buildfile.m — Define TestTask with source and report options
2. Add CI config — GitHub Actions, Azure DevOps, or GitLab CI template

Running Tests

Via MCP

Use the run_matlab_test_file MCP tool to run test files directly. When you need filtering, parallel execution, or other options, use evaluate_matlab_code with runtests:

matlabCopy code
results = runtests('tests');                            % all tests in folder
results = runtests('tests', Tag='Unit');                % by tag
results = runtests('tests', Name='*Calculator*');       % by name pattern
results = runtests('tests', UseParallel=true);          % parallel execution
results = runtests('tests', Strict=true);               % warnings = failures

Analyzing Results

matlabCopy code
disp(results);

for r = results([results.Failed])
    fprintf('\nFAILED: %s\n', r.Name);
    disp(r.Details.DiagnosticRecord.Report);
end

for r = results([results.Incomplete])
    fprintf('\nINCOMPLETE: %s\n', r.Name);
    disp(r.Details.DiagnosticRecord.Report);
end

Coverage Analysis

matlabCopy code
import matlab.unittest.TestRunner
import matlab.unittest.plugins.CodeCoveragePlugin
import matlab.unittest.plugins.codecoverage.CoverageResult

runner = TestRunner.withTextOutput;
covFormat = CoverageResult;
runner.addPlugin(CodeCoveragePlugin.forFolder('src', ...
    IncludingSubfolders=true, Producing=covFormat, MetricLevel='mcdc'));
results = runner.run(testsuite('tests'));

covResults = covFormat.Result;
summary = coverageSummary(covResults, "mcdc");
fprintf('Coverage: %d/%d (%.1f%%)\n', sum(summary(:,1)), sum(summary(:,2)), ...
    100*sum(summary(:,1))/sum(summary(:,2)));

For detailed coverage workflows — gap analysis, justifications, generated C code coverage, and data type/size coverage — see references/code-coverage-guidance.md.

Key Functions

CI/CD Integration

Use buildtool with a buildfile.m for CI pipelines. See references/test-execution-guidance.md for buildfile.m templates and CI configs (GitHub Actions, Azure DevOps, GitLab CI).

References

Load these on demand — most test runs only need what's in this file.

Conventions

• Always: run tests via the run_matlab_test_file MCP tool when possible
• Always: report pass/fail summary before taking further action
• Always: check for Failed/Incomplete tests before analyzing coverage
• Prefer: forFolder with IncludingSubfolders=true as the default source specification for coverage
• Prefer: MC/DC as the default metric level because each level includes all lower levels
• Never: generate coverage reports (HTML, Cobertura) unless the user explicitly requests them
• Never: use CoverageResult as the only format — always use it as the primary programmatic format

Copyright 2026 The MathWorks, Inc.