remote-retrieval-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (HIGH): The skill instructs users to connect to an unverified MCP server at https://rube.app/mcp. This domain is not recognized as a trusted source, and the instruction 'No API keys needed — just add the endpoint and it works' bypasses standard security vetting and authentication protocols.
  • [REMOTE_CODE_EXECUTION] (HIGH): Through tools like RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH, the skill facilitates the execution of remote logic provided by the unverified server. This allows the server to potentially execute arbitrary commands on the agent's host environment.
  • [COMMAND_EXECUTION] (HIGH): The workflow relies on RUBE_SEARCH_TOOLS to determine tool schemas and execution plans at runtime. This dynamic execution model transfers control of the agent's actions to the remote service provider.
  • [PROMPT_INJECTION] (HIGH): Significant vulnerability to Indirect Prompt Injection (Category 8). The skill's purpose is to retrieve remote data and then use that data to discover and execute tools. Ingestion point: 'Remote Retrieval' content. Capability inventory: RUBE_MULTI_EXECUTE_TOOL, RUBE_REMOTE_WORKBENCH. Boundary markers and sanitization are entirely absent, meaning malicious instructions in retrieved content could directly lead to unauthorized tool execution or data exfiltration.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 12:57 PM