linear-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: 0.80). The skill requires connecting to an MCP server at runtime (via RUBE_SEARCH_TOOLS) and explicitly names https://rube.app/mcp as the MCP endpoint, which serves tool schemas that directly determine agent instructions/behavior, so this external URL can control prompts at runtime.
Audit Metadata