dock-certs-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • REMOTE_CODE_EXECUTION (HIGH): The skill utilizes RUBE_REMOTE_WORKBENCH and RUBE_MULTI_EXECUTE_TOOL which facilitate execution of operations in a remote environment provided by an untrusted domain (rube.app).
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill requires adding an external, non-whitelisted MCP endpoint (https://rube.app/mcp), creating a dependency on unverified infrastructure.
  • PROMPT_INJECTION (HIGH): High vulnerability to Indirect Prompt Injection (Category 8).
  • Ingestion points: Tool schemas and 'recommended execution plans' are fetched from RUBE_SEARCH_TOOLS at runtime from a non-whitelisted source.
  • Boundary markers: Absent; instructions mandate following the remote server's output and 'recommended execution plans' without local validation.
  • Capability inventory: RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH provide substantial write/execute capabilities.
  • Sanitization: Absent; the agent is instructed to use schemas exactly as returned by the remote server.
  • COMMAND_EXECUTION (HIGH): The instruction to follow 'execution plans' and tool slugs provided by a remote server at runtime constitutes a bypass of local security constraints, granting the remote server control over the agent's actions.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 12:33 PM