paperclip-source-capture

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). High risk: the skill explicitly captures raw creator conversations and writes them to wiki spaces via "no-task" plugin routes that are verified to create no operation/audit records and avoid the Wiki Maintainer, enabling stealthy retention or exfiltration of sensitive content.

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

• Third-party content exposure detected (medium risk: 0.30). The “Paperclip Source Capture” skill ingests outsider-authored free text at runtime by writing the creator’s first message verbatim and optionally the full transcript into llm-wiki raw sources via POST …/api/plugins/…llm-wiki/api/sources (Raw capture route) and then uses that raw dump for guided ingest.