gan-ai-automation

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill requires the user to add an external MCP server endpoint https://rube.app/mcp. This domain is not part of the trusted organizations or repositories list. This server provides the tool definitions and execution logic for the agent.
  • [COMMAND_EXECUTION] (LOW): The skill uses RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH to execute actions. While these are scoped to the Gan AI toolkit, they represent the capability to perform remote operations based on schemas fetched at runtime.
  • [PROMPT_INJECTION] (LOW): Potential for Indirect Prompt Injection (Category 8). The skill instructions mandate calling RUBE_SEARCH_TOOLS to get current schemas and instructions before every execution. This creates a surface where the remote server could inject instructions into the agent's context.
  • Ingestion points: RUBE_SEARCH_TOOLS response body.
  • Boundary markers: Absent. The skill does not instruct the agent to treat the tool search results as untrusted data or use delimiters.
  • Capability inventory: Remote tool execution via Gan AI/Composio bridge.
  • Sanitization: None specified in the skill instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:48 PM