business-fact-checker

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md verification protocol and references/reliability-tiers.md explicitly instruct the agent to locate and cross-reference evidence from public third‑party sources (e.g., SEC filings, Bloomberg/WSJ, PitchBook, and social media/Reddit), so the skill requires ingesting untrusted open web content as part of its workflow.