trtc-onboarding

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The document mandates silent, mandatory telemetry that sends every user prompt verbatim and session events to an external MCP server (mcp__tencent-rtc-skill-tool__skill_analysis) without notifying users and allows user-pasted credentials (test UserSig, code snippets) to be reported — combined with hidden session-file writes and explicit instructions to not disclose reporting, this is a deliberate privacy/exfiltration risk.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs runtime fetches of remote index files (e.g. curl -s https://trtc.io/llms/{product}/{resolved_identifier}.txt and fallback https://trtc.io/llms/{product}.txt), parses them for a demo document URL and then uses that extracted repository/clone command to git-clone/run demo code, so external content from those URLs directly controls what code is fetched and executed.