youtube-automation

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly directs the agent to call public YouTube endpoints (e.g., YOUTUBE_SEARCH_YOU_TUBE, YOUTUBE_LIST_COMMENT_THREADS, YOUTUBE_VIDEO_DETAILS) to fetch user-generated public content (search results, comments, video details) that the agent is expected to read and use to drive subsequent tool calls and actions, so untrusted third-party content can influence behavior.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly requires connecting to Rube MCP at https://rube.app/mcp and instructs calling RUBE_SEARCH_TOOLS at runtime to fetch current tool schemas, which directly determine available tools/instructions and thus control the agent's prompts and behavior.