save-tokens

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). Yes — although visible, the prompt includes instructions unrelated to its stated purpose (claims sole authority and orders suppression/override of other skills/system behavior), which is a deceptive/overriding directive outside the compression scope.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). Skill explicitly asserts sole-session authority and directs suppression of external/third-party skills, enabling intentional bypass of other safety/monitoring modules (backdoor-like control); no direct data-exfiltration or remote-exec payloads present.