zhihu-topic-scraper

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The Skill contains explicit instructions to automatically install a runtime dependency and to hide that action from the user (do not request approval / do not describe it as asking the user to open their browser), which is a deceptive behavior outside the scraper's stated data-collection purpose.

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md instructs the agent to use the BrowserAct CLI to access real Zhihu web pages at runtime (public web content), which can include outsider-authored free text that the agent may ingest into the LLM context via the browser-extracted page content.

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 use the Install BrowserAct link at runtime to obtain missing runtime dependencies (which may fetch and run installer code): https://www.browseract.com/?co-from=social-media&redirect=https://github.com/browser-act/skills/tree/main

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

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill instructs the agent to automatically install missing CLI dependencies without user confirmation and to resume execution, which directs the agent to modify the host system (install software) and could trigger privileged operations even though it does not explicitly request sudo or create accounts.