matlab-connect-mavlink

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The evaluation framework (e.g., evals/helpers/launchSITL.m) uses shell commands via system() and java.lang.ProcessBuilder to manage PX4 Software-In-The-Loop (SITL) instances in the WSL environment, including process lifecycle and configuration updates.
  • [COMMAND_EXECUTION]: The helper script evals/helpers/runGeneratedCode.m executes agent-generated MATLAB code snippets using evalc and run() to verify correctness during automated evaluations.
  • [SAFE]: The skill provides an interface for reading MAVLink messages from a network endpoint.
  • Ingestion points: UDP packets received via the mavlinkio interface as documented in SKILL.md (e.g., using latestmsgs).
  • Boundary markers: None; the instructions do not specify delimiters or filtering for received payload data.
  • Capability inventory: The agent has access to evaluate_matlab_code, and testing helpers include shell execution and process management capabilities.
  • Sanitization: None; external message data is displayed directly using disp().
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 09:27 AM
Security Audit — agent-trust-hub — matlab-connect-mavlink