mc-case-studies
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The analyzed files (SKILL.md and reference.md) are strictly instructional and descriptive markdown documents. They define a framework for music analysis but do not include any shell commands, scripts, or executable code.- [INDIRECT_PROMPT_INJECTION]: The skill mentions processing external audio data (multitracks from the Cambridge-MT dataset) using local repository tools. While this constitutes a data ingestion surface, the processed data is converted into numerical parameters for musicological analysis rather than natural language instructions, presenting no actionable attack surface for the agent. No sanitization or boundary markers are explicitly defined for this data processing pipeline, but the risk remains negligible given the primary purpose of the skill.
Audit Metadata