dromo-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill possesses a high-risk attack surface where external content from the rube.app MCP server and Dromo API influences agent actions.
  • Ingestion points: Data returned from RUBE_SEARCH_TOOLS and individual tool outputs.
  • Boundary markers: None present to distinguish instructions from data.
  • Capability inventory: Write and execute capabilities via RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH.
  • Sanitization: No evidence of input validation or escaping.
  • [Unverifiable Dependencies] (MEDIUM): Requires configuration of an external MCP server from an unverified source (https://rube.app/mcp), which serves as a runtime extension.
  • [Dynamic Execution] (MEDIUM): Tool selection and argument construction are performed dynamically based on remote search results (RUBE_SEARCH_TOOLS).
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 12:33 PM