reply-automation
Warn
Audited by Snyk on Feb 16, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). The skill requires adding and querying the runtime MCP endpoint https://rube.app/mcp so RUBE_SEARCH_TOOLS/RUBE_MANAGE_CONNECTIONS fetch tool schemas and execution plans at runtime, meaning remote content from that URL directly controls agent instructions and tool execution.
Audit Metadata