Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 87% confidence
- Finding
- The skill explicitly instructs the agent to inspect repositories, implement code directly, and work with Discord bots, dashboards, OAuth flows, webhooks, deployment, and worker systems. Those tasks inherently imply network and environment-variable access, yet the skill declares no permissions, creating a mismatch that can bypass least-privilege review and lead to unintended secret exposure or outbound actions when the skill is invoked.
