topic-resolver

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 resolves a user-provided topic into entities and then triggers searches/deep crawls, which can fetch outsider-authored web content (e.g., docs/blog pages, GitHub releases, Reddit posts) and feed that readable text into the agent’s LLM context via the subsequent search/crawl steps.