computer-use

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

• Third-party content exposure detected (low risk: 0.10). The agent operates browsers and desktop GUIs where arbitrary web pages can be navigated and read, introducing a low risk of indirect prompt injection from fetched web content.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the execution of external tools from third-party repositories via npx and uvx (such as web-infra-dev/midscene-skills and am-will/codex-skills), presenting unverifiable community runtime dependencies under the high tier.