legal-guidance-vault

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow can ingest outsider-authored free text from Slack threads (and potentially meeting transcripts/notes if they include third-party speech) via the “Archiving from messages” path: when the user provides a Slack thread or asks to archive/scan flagged messages, the skill fetches the full thread content and extracts it into the LLM context.