simulink-optimizing-embedded-code
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONOBFUSCATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a 'Customer Extensions' protocol that automatically discovers and ingests instructions from external files within the project repository.
- Ingestion points: The agent reads 'optimization_preferences.yaml' and 'reference.md' files from the '.custom_optimizations/' directory (documented in 'assets/customer-extensions.md').
- Boundary markers: Absent; the skill is instructed to semantically match plain-English rules from these files against its own internal logic.
- Capability inventory: The skill can execute shell commands via 'system()' and run complex logic through MATLAB scripts and functions.
- Sanitization: No specific sanitization or escaping of this external content is described before it is used to guide the optimization workflow.
- [COMMAND_EXECUTION]: The main 'SKILL.md' instructions direct the agent to execute shell commands using the MATLAB 'system' and 'evalc' functions to interact with the Git version control system (e.g., 'git rev-parse HEAD' and 'git status --porcelain') for project state tracking.
- [OBFUSCATION]: The skill includes several MATLAB P-files ('.p') in the 'scripts/' directory and an encrypted file ('assets/ecokg.enc'). These are protected binary formats used by the author (MathWorks) to distribute proprietary logic and data, making them opaque to static analysis.
Audit Metadata