figma-to-code

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The "Companion check" tells the agent to run shell commands (ls) against local skill directories and to offer/install other skills with npx, which is an instruction to access and modify the host environment that is unrelated to the figma-to-code purpose and thus a hidden/deceptive action outside the skill's stated scope.

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

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill explicitly instructs the agent to run local shell commands (ls) to inspect user skill directories and suggests installing missing companions via an npx install command, which involves accessing the filesystem and potentially modifying the machine state (installing packages), so it pushes actions that can change the host environment even though it doesn't request sudo or privileged system changes.