matlab-import-export-vehicle-data

Pass

Audited by Gen Agent Trust Hub on Aug 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted vehicle network database and log files which serves as a potential surface for indirect prompt injection attacks.
  • Ingestion points: The skill uses fileread() in SKILL.md, mdfRead() in references/mdf-import.md, blfread() in references/blf-import.md, and canSignalImport() in references/asc-import.md to ingest external data.
  • Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following potential instructions embedded within the metadata or signal names of the processed files.
  • Capability inventory: The skill utilizes the evaluate_matlab_code tool as specified in manifest.yaml, which executes MATLAB code generated or influenced by the processed data.
  • Sanitization: No sanitization or validation logic is present to filter signal names, message names, or comments from the external files before they are interpolated into the agent's context or code.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 2, 2026, 12:12 PM
Security Audit — agent-trust-hub — matlab-import-export-vehicle-data