literature-review

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). The list contains one high-risk entry: a direct shell installer URL (https://parallel.ai/install.sh) which encourages running a downloaded .sh via curl|bash from a third-party domain; the rest are academic, documentation, or API links and are low risk.

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

• Third-party content exposure detected (medium risk: 0.30). In the required runtime workflow, outsider-provided search/extraction queries lead the agent to ingest free text from web search/extracted academic paper content via parallel-cli search and parallel-cli extract, which are used without selecting a specific trusted item first.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill requires executing remote installation code via "curl -fsSL https://parallel.ai/install.sh | bash" (SKILL.md), which fetches and runs remote code as a required dependency.