hada-scout

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and parses the public hada.io RSS feed (https://feeds.feedburner.com/geeknews-feed) in Phase 1 and feeds those item titles/URLs into an LLM pre-scout and downstream /scout tool invocations, so untrusted third-party feed content can directly influence scoring and subsequent automated actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches https://feeds.feedburner.com/geeknews-feed at runtime and injects the fetched item titles/URLs into the haiku agent prompt (the {numbered_item_list}), meaning remote feed content directly controls model inputs used for relevance decisions.