kimi-webbridge

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to navigate to arbitrary websites and read/interpret page content using tools like "navigate" and "snapshot" (SKILL.md), so it ingests untrusted public web content that can materially influence subsequent clicks/fills/evaluate actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's operations instructions tell the user to run a runtime install command that fetches and executes remote code via curl -fsSL https://kimi-web-img.moonshot.cn/webbridge/install.sh | bash, which directly executes external code the skill depends on.