Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 88% confidence
- Finding
- The skill instructs the agent to detect the OS, read reference files, and run platform-specific cleaning scripts, which implies shell, file-read, and environment access without declaring corresponding permissions. Undeclared capabilities are dangerous because they hide the true execution surface from reviewers and users, especially for a destructive disk-cleaning skill that can remove files across the system.
