habitica-automation

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill instructs users to add an external MCP endpoint https://rube.app/mcp. This domain is not a trusted source, meaning the agent's tool definitions are controlled by an unverified third party.
  • REMOTE_CODE_EXECUTION (MEDIUM): The presence of RUBE_REMOTE_WORKBENCH and RUBE_MULTI_EXECUTE_TOOL indicates that the skill can execute complex logic or tools remotely. Since these tools are fetched dynamically at runtime via RUBE_SEARCH_TOOLS, the agent is executing logic whose behavior is not statically defined in the skill itself.
  • INDIRECT PROMPT INJECTION (LOW): The skill is highly susceptible to indirect injection because it explicitly follows 'recommended execution plans' and 'tool schemas' fetched from the remote server. \n
  • Ingestion points: Tool definitions and plans returned by RUBE_SEARCH_TOOLS.\n
  • Boundary markers: Absent; the skill encourages exact adherence to remote results.\n
  • Capability inventory: Remote tool execution and workbench access via RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH.\n
  • Sanitization: Absent; the skill lacks validation for the data returned by the search tool.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 05:45 PM