Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 92% confidence
- Finding
- The skill documentation directs the agent to read local configuration files and execute shell scripts, but it declares no corresponding permissions. This creates a mismatch between advertised and actual capabilities, which can cause the skill to run with broader implicit access than reviewers or policy enforcement expect. In this context, the skill also sets up SSH-based wrappers to a remote Mac and can influence command execution paths, making undeclared file-read and shell access more sensitive than a purely local utility skill.
