cubox

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

• Third-party content exposure detected (low risk: 0.10). The Cubox CLI skill reads user-saved bookmarks, web pages, and articles via the local cubox-cli tool, which fetches personal reading memory that can contain arbitrary outsider-authored text (such as saved web articles).

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

• Potentially malicious external URL detected (medium risk: 0.30). The skill instructs the agent to display and reference an update command that uses npx skills add OLCUBO/cubox-cli -g -y, which is a remote GitHub repository locator fetched and executed via a package runner during an update workflow.