feed-wsj

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The prompt overtly includes actionable, paywall-bypass instructions for fetching full WSJ article bodies—even after stating the adapter is "headlines only" and "Do NOT scrape the full body"—which is a deceptive instruction that changes behavior outside the skill's stated purpose.

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow fetches WSJ RSS items from feeds.a.dj.com and ingests the RSS title/link/pubDate (and possibly description teaser) which are publisher-provided text from an external third party (WSJ), i.e., outsider-authored free text enters the LLM context.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill explicitly advises bypassing paywalls (e.g., using the "bypass-paywalls-clean" Chrome approach that clears metered cookies/disables paywall JS) which instructs circumventing a security/ToS mechanism, even though it doesn't request sudo, system-file changes, or user-account creation.