Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 89% confidence
- Finding
- The skill declares no permissions, but its documented behavior and usage clearly require reading user-provided genetic files and writing a report. This creates a security and governance gap: operators may approve or sandbox the skill under incorrect assumptions, while the runtime still handles sensitive genomic data and writes output files. In this context, the issue is more serious because the input is highly sensitive health/genetic information, even though file I/O is expected for the skill’s function.
