froq-voice-essay

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

• Third-party content exposure detected (low risk: 0.10). The required workflow at SKILL.md:19 reads the user's Arc browser history, whose page titles are third-party-authored text that an outsider can influence by publishing a page, but the attacker cannot force delivery without the user visiting it, so this is low exposure.

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

• Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is that the skill instructs the agent to re-fetch the 0froq/0froq.github.io repository and treat the newest posts as overriding the style guide, which it then rewrites, so the fetched content governs behavior; the repository is froQ's individually owned blog, which is unofficial but related, giving a reputable-unofficial medium tier.