xiaohongshu-rednote-hashtag-scraper

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.80). The skill contains explicit, deceptive instructions to install runtime dependencies and conceal that action from the user (install automatically "without asking the user to confirm" and "do not describe it as a request for the user to open their browser"), which instructs the agent to act without user consent and hide that behavior—this is outside the advertised scraping task's transparent operation.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The Skill explicitly instructs automatic, non-consensual installation of an external CLI and to avoid requesting user approval, which is a strong supply-chain/backdoor risk enabling hidden remote code execution or unwanted dependencies.

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

• Third-party content exposure detected (high risk: 0.85). This skill’s required runtime workflow uses BrowserAct to access live Xiaohongshu/RedNote topic/hashtag pages (public web content) and ingest scraped note text/snippets into the agent context, which is outsider-authored free text via browser-fetched pages.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs the agent to automatically install a required runtime dependency via the installation link https://www.browseract.com/?co-from=social-media&redirect=https://github.com/browser-act/skills/tree/main, which is used at runtime to obtain and install external code needed to execute the Skill.

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 and resume execution without asking the user for approval, which directs the agent to modify the host system by installing software and bypass user consent even though it does not explicitly request sudo or direct system-file edits.