update

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md requires syncing/decoding tasks from external sources (e.g., GitHub Issues and “MCP” project trackers, plus “deep scan” of chat/email/calendar/docs in --comprehensive), which would ingest outsider-authored free text (issue bodies, chat/email messages, etc.) into the agent’s LLM context for diffing, triage, and memory extraction.