design-shotgun

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill is a legitimate design-mockup generator, but it tells the agent to pass mode: "bypassPermissions" on subagent calls and to run unattended, which weakens permission protections beyond what generating HTML mockups requires, though it is disclosed and not aimed at deceiving the user.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.65). The skill is otherwise a benign design-mockup generator, but its Permission Mode section tells the agent to pass mode: "bypassPermissions" on subagent calls and to verify the effective mode for unattended runs, which bypasses the permission safeguards that protect the machine.