social-intent-monitoring

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

• Third-party content exposure detected (high risk: 0.85). The required workflow described in SKILL.md involves monitoring public social conversations (e.g., LinkedIn/X/Reddit posts and comments) and using those signal texts as “signal context” for downstream enrichment/outreach, meaning outsider-authored free-form web/social content would be ingested into the agent’s LLM context at runtime.