Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 96% confidence
- Finding
- The skill instructs the agent to run Python scripts, inspect local files, glob input datasets, and write outputs, which clearly implies shell execution plus file read/write capabilities without any declared permission boundary. That is dangerous because users and hosting platforms cannot accurately assess or constrain what the skill may access, especially when it processes arbitrary paths and large datasets.
