k-skill-cleaner

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill scans local coding-agent transcripts and JSONL logs that contain user and agent dialogue/session history, which can include outsider-authored content or prompt injections.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs downloading and executing external packages via npx -y @nomadamas/k-skill@0 from npm/GitHub during runtime, presenting a weak or unofficial third-party operational dependency.