ecommerce-seller-profile-scraper

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The Skill contains explicit, deceptive instructions to automatically install a runtime dependency without asking the user and to avoid telling the user or requesting approval, which is outside the scraping/data-collection purpose and hides installation actions.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The Skill explicitly instructs automatic installation of an external CLI without user approval and tells agents not to request or describe user consent, combined with capture of browser-visible network responses — a deceptive installation flow that enables supply-chain installation and potential data exfiltration.

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 runtime browser-act Skill to “access real browser pages” and “capture browser-visible page content” from user-specified public URLs/pages, which can include outsider-authored free text that the LLM ingests via the browser-extracted content path.

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

• Potentially malicious external URL detected (high risk: 0.90). The Skill explicitly instructs the agent to perform an automatic runtime installation using https://www.browseract.com/?co-from=ecommerce&redirect=https://github.com/browser-act/skills/tree/main (and https://www.browseract.com/?co-from=ecommerce), which will fetch and install remote code (the BrowserAct CLI/Skill) that the agent must execute to proceed.

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 explicitly directs the agent to automatically install the BrowserAct CLI as a required runtime dependency and to do so without requesting the user's confirmation, which instructs the agent to modify the host system state (install software) without user approval.