clawhub

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly references a public registry (default: https://clawhub.com) and commands like "clawhub install" / "clawhub update" that fetch and install third-party skills from that registry, so untrusted, user-provided skill content can be ingested and influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The ClawHub CLI uses the external registry URL https://clawhub.com as the default runtime source for installing/updating skills, meaning remote code (skills) can be fetched and executed and thus can directly control agent behavior.