Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 91% confidence
- Finding
- The skill explicitly instructs the agent to execute a shell command and to read/write accounting artifacts (`rapprochement.json`, batch reports), yet it declares no permissions. This creates a capability/permission mismatch that can bypass expected trust boundaries, making it harder for reviewers and runtime policy to understand or constrain file-system and command execution behavior.
