douyin-video-extractor

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt shows and instructs embedding API keys directly (env/API_KEY, CLI --api-key, and MCP JSON env entries), which would require the model to include secret values verbatim in generated commands/configs and thus risks secret exfiltration.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). These URLs mix legitimate services (Douyin short links, GitHub, modelcontextprotocol.io, ara.so) with risky patterns — an unknown GitHub repo, third‑party API hosting (cloud.siliconflow.cn), and a direct curl|sh installer (https://astral.sh/uv/install.sh) plus multiple short/redirecting links — making it a moderate-to-high risk for distributing malware if you execute downloaded scripts or binaries without review.

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime path for outsider text is via Douyin share links: the server fetches Douyin video pages/metadata and then ingests the resulting page-derived text (e.g., title/caption/transcript-related content) into the LLM context for transcription, which is outsider-authored web content.

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

• Potentially malicious external URL detected (high risk: 0.90). The installation instructions explicitly run a remote install script (curl -LsSf https://astral.sh/uv/install.sh | sh) and require cloning the repository (https://github.com/yzfly/douyin-mcp-server.git) which are fetched at setup and lead to executing remote code from those URLs.