to-video

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md’s process step 3, the skill passes user-provided planning artifacts and related text into the selected HyperFrames workflow as “project truth,” meaning outsider-authored free text can be ingested by the runtime workflow without first selecting a specific pre-approved item.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly loads the runtime endpoint "/hyperframes" to let it select a workflow and read its route contract (and also calls "/listenhub-tts" and "/media-use"), which means external runtime content will directly control the agent's workflow/instructions and produce execution outputs.