test-browser

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts a "live URL" as a required input and the runbook/CLI steps (references/agent-browser-runbook.md and SKILL.md) instruct the agent to open arbitrary URLs (agent-browser open / open "http://.../") and verify page content, meaning the agent will fetch and interpret untrusted public web content as part of its workflow.

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

• Potentially malicious external URL detected (high risk: 0.75). The runbook explicitly directs runtime installation/use of an external CLI via "npm install -g agent-browser" (fetching from the npm registry, e.g. https://registry.npmjs.org/), which would download and install remote code that the skill then executes and relies on for browser operations.