ably-automation

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

• Third-party content exposure detected (high risk: 0.80). The skill requires calling the external Rube MCP endpoint (https://rube.app/mcp) via RUBE_SEARCH_TOOLS to fetch tool slugs and schemas from that third-party server, and those returned, untrusted tool descriptions are explicitly read and used to decide which tools/arguments to execute, allowing indirect injection of instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). Flagging https://rube.app/mcp because the skill requires adding that MCP server and at runtime calls RUBE_SEARCH_TOOLS / RUBE_MANAGE_CONNECTIONS to fetch tool schemas and recommended execution plans, which directly control agent prompts and tool execution.