fingertip-automation

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The skill requires adding an external MCP server (rube.app/mcp). While this is necessary for functionality, the domain is not a pre-approved trusted source.
  • REMOTE_CODE_EXECUTION (LOW): The inclusion of RUBE_REMOTE_WORKBENCH and RUBE_MULTI_EXECUTE_TOOL implies the execution of operations in a remote environment managed by the third-party service.
  • PROMPT_INJECTION (LOW): The skill is vulnerable to indirect prompt injection through its dynamic tool discovery mechanism.
  • Ingestion points: Tool schemas and execution plans are fetched at runtime from the Rube MCP server via RUBE_SEARCH_TOOLS.
  • Boundary markers: None present; the agent is instructed to use the exact field names and types from search results without validation.
  • Capability inventory: The skill possesses the ability to execute multiple tools and access a remote workbench.
  • Sanitization: No sanitization or validation of the fetched metadata is performed, meaning malicious tool schemas from the server could influence agent behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:46 PM