social-intelligence

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

• Third-party content exposure detected (high risk: 0.95). The skill’s knowledge extraction path (/extract → scripts/engagement-loop.py extract → get_comments(pid) / content = c.get("content")) ingests Moltbook comment body text authored by other users (outsider) and then writes it into LLM-readable markdown notes (research/notes/...-social-insights-raw.md) for later LLM synthesis.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill explicitly instructs bypassing X's API-level 403 protections by using a Playwright-driven headless Chromium profile and manual app actions (an API bypass), which is a directive to evade a platform security mechanism even though it does not request sudo, create users, or modify system-level files.