afa-product

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

• Third-party content exposure detected (high risk: 1.00). The skill's core workflow (e.g., Mode B Tier 3 "竞品痛点逆向工程" in SKILL.md) and supporting references (references/core-frameworks.md and references/differentiation-playbook.md) explicitly instruct collecting and analyzing user-generated public content (e.g., Amazon/Reddit/YouTube reviews and comments), which the agent is expected to read and use to drive product decisions, creating a clear path for indirect prompt injection.