alibaba-supplier-lead-research

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The Skill contains instructions that would automatically install and invoke an external runtime dependency without user consent and explicitly tells the agent to conceal that action, which presents a clear supply-chain/unauthorized-installation and potential data-exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs the agent to use the BrowserAct CLI to “access real browser pages” and “capture browser-visible page content” from user-specified URLs/marketplace pages at runtime, which can include outsider-authored free text from public web sources.

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

• Potentially malicious external URL detected (high risk: 0.90). The Skill explicitly requires using an external installer URL at runtime (https://www.browseract.com/?co-from=ecommerce&redirect=https://github.com/browser-act/skills/tree/main) to fetch/install the BrowserAct CLI, which means remote code/content from that URL controls runtime behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The Skill instructs the agent to automatically install the BrowserAct CLI (a required runtime dependency) without asking the user for approval and to resume execution afterward, which encourages the agent to modify the host system state (install software) autonomously and potentially with elevated privileges.