matlab-run-tests
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the
evaluate_matlab_codeandrun_matlab_test_filetools to execute MATLAB code blocks and test files. This is the core functionality for running test suites and coverage plugins. - [DYNAMIC_EXECUTION]: The skill instructions include patterns for generating XML filter files at runtime using
writelines. These files are used to justify uncovered code branches during coverage analysis. This represents low-risk dynamic content generation from internal templates. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze potentially untrusted data from test execution outputs.
- Ingestion points: Test diagnostic records (
r.Details.DiagnosticRecord.Report) and source code files instrumented for coverage. - Boundary markers: The instructions do not define specific delimiters or "ignore" instructions when processing failed test diagnostics.
- Capability inventory: The agent has access to
evaluate_matlab_codeandrun_matlab_test_file, which allow execution of arbitrary MATLAB commands. - Sanitization: There are no explicit instructions to sanitize or escape diagnostic reports before the agent processes them, creating a surface where a malicious test failure could influence agent behavior.
Audit Metadata