slci-inspection-incompatibility-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the evaluate_matlab_code tool to execute MATLAB functions like rtwbuild, slbuild, and slci.Configuration. These commands are used to regenerate code and manage the inspection process within the user's MATLAB environment as described in SKILL.md.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external SLCI inspection reports (HTML, PDF, or DOC formats) and generated C source files to identify failure patterns.
- Ingestion points: Workflow Step 2 (Parse the Inspection Report) and Step 3 (Read the Failed Code Lines) in SKILL.md.
- Boundary markers: The skill does not use specific delimiters but enforces a confirmation step (Step 0) where the user must approve the diagnostic action before the agent accesses files.
- Capability inventory: Uses evaluate_matlab_code in SKILL.md to run build and inspection commands.
- Sanitization: No explicit sanitization or filtering of the report content or source code content is performed during parsing.
Audit Metadata