verasic-deep-research

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). One URL is high-risk: the raw.githubusercontent.com setup.sh script is referenced and the README shows it being piped to bash (curl | bash), which is a common and dangerous pattern for distributing malware if the script is untrusted.

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

• Third-party content exposure detected (high risk: 0.85). The workflow explicitly performs runtime web fetching/extraction (e.g., fetch-url / extract-excerpt) of non-user-chosen public pages, then ingests those extracted snippets/bodies into the agent’s LLM context for subsequent verification/synthesis—an outsider-authored free text path via the HTTP/browser fetch and extract steps.

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

• Potentially malicious external URL detected (high risk: 1.00). The README includes an installation command that fetches and pipes a remote script to bash (https://raw.githubusercontent.com/Milkywayrules/verasic-skills/main/setup.sh), which executes remote code during setup.