parallel-web-extract

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). docs.parallel.ai looks like a documentation site (low risk) but parallel.ai/install.sh is a direct remote shell script intended to be downloaded and executed — delivering and running an unverified .sh via curl|bash is a high‑risk distribution vector that could carry malware unless the domain and script are independently verified and cryptographically signed.

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly directs the skill to fetch arbitrary public URLs using parallel-cli extract ("Extract content from: $ARGUMENTS" and examples like extracting webpages/articles/PDFs), returning verbatim third‑party content that the agent will read and could use to influence follow-up actions, so untrusted web content could enable indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's setup instructs executing a remote install script at runtime via "curl -fsSL https://parallel.ai/install.sh | bash", which fetches and runs external code that the skill may rely on, so this URL is a high-risk runtime dependency.