ai-audio-generation

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). These URLs point to an unverified personal service (luxin.sh and its api subdomain) that is referenced as the canonical runtime/CLI provider for the skill and is used to deliver executable guidance (e.g., unpinned npx installs and service-supplied commands), making the domain a suspicious source for distributing malware.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill contains instructions to download/execute unpinned remote packages (npx @latest / -y / unversioned installs) and to run a command string supplied by the hosted service (data.next_command), creating high-risk supply-chain and remote code-execution/backdoor vectors.

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

• Third-party content exposure detected (medium risk: 0.30). The Luxin runtime workflow reads and executes a “guide” response via data.next_command (hosted service-controlled command/fields), so outsider-authored free text could be injected into that guide content path and be ingested before selection of a specific item.

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

• Potentially malicious external URL detected (high risk: 0.96). The skill instructs agents to execute service-supplied commands (data.next_command) from Luxin's hosted runtime, which is served at https://api.luxin.sh and therefore can deliver runtime instructions that control execution.