ln-502-regression-checker

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.70). The prompt explicitly claims it will "Never modifies Linear or kanban" but repeatedly includes instructions to "add Linear comment" and "Linear comment posted," which is a deceptive/contradictory instruction that changes state outside the skill's stated scope.

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill forces the agent to read and reuse exact commands and environment-variable settings from the runbook (including Docker envs), which can require echoing secret values verbatim into commands or outputs, creating an exfiltration risk.