tam-sam-som-builder

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

• Third-party content exposure detected (medium risk: 0.65). The skill ingests memory/market_insights/<niche>-*-<YYYY>-<MM>.md at runtime and uses their YAML frontmatter and narrative (e.g., “Platform narrative (Reddit pain points, TikTok engagement, App Store reviews)”), which is outsider-authored free text if those trend files were populated from public web/community sources; this text is then placed into the agent’s LLM context for extraction and sizing.