cybersecurity

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.85). The skill explicitly requires dispatching the user's "verbatim task" to sub-agents and relaying sub-agent outputs verbatim, which forces the LLM to include any user-supplied secrets (API keys, tokens, passwords) in generated messages if they appear in the task or sub-agent output.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Most URLs are benign docs, vendor pages, and OSS repos, but several entries are attacker-controlled or malformed (attacker.com, example.com.attacker.com, victim.com@attacker.com, open-redirect payloads) and therefore look suspicious and could be used to host or redirect to malware/phishing content.