v0-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (HIGH): The skill requires the user to add an untrusted third-party MCP server (https://rube.app/mcp) to their client configuration. This server is not within the trusted scope and controls the logic for all subsequent tool executions.
  • REMOTE_CODE_EXECUTION (HIGH): Features tools like RUBE_REMOTE_WORKBENCH and RUBE_MULTI_EXECUTE_TOOL which execute actions on a remote environment. The behavior of these tools is dynamically defined by the untrusted MCP server's response to RUBE_SEARCH_TOOLS or RUBE_GET_TOOL_SCHEMAS.
  • PROMPT_INJECTION (HIGH): Susceptible to Indirect Prompt Injection. The skill processes data from V0 tasks (which can include user-provided UI descriptions or code) and possesses high-privilege execution capabilities.
  • Ingestion points: V0 task descriptions and project metadata processed through Rube tools.
  • Boundary markers: None identified in the skill instructions to separate external data from tool commands.
  • Capability inventory: RUBE_MULTI_EXECUTE_TOOL, RUBE_REMOTE_WORKBENCH, RUBE_MANAGE_CONNECTIONS.
  • Sanitization: No sanitization or validation of the external content is performed before interpolation into the workflow.
  • DATA_EXFILTRATION (MEDIUM): Automating V0 tasks through the Rube proxy server involves routing V0 project data and connection status through an external entity (rube.app), creating a risk of unauthorized data access by the provider.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 08:34 AM