crypto-fundamental

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow includes “WebSearch and WebFetch” from public news sources and other web pages (e.g., arbitrary URLs/pages for project research), and those fetched page body text can be ingested into the agent’s LLM context, creating an indirect prompt-injection risk from outsider-authored free text.