waboxapp-automation
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS] (SAFE): The skill requires the user to add an MCP server endpoint (https://rube.app/mcp). This is a standard configuration for the Model Context Protocol and does not involve automated or hidden downloads of executable scripts.
- [COMMAND_EXECUTION] (SAFE): The skill utilizes RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH. These are structured tool-calling commands rather than arbitrary shell command execution.
- [PROMPT_INJECTION] (SAFE): No evidence of jailbreak attempts, system prompt extraction, or safety filter bypass instructions was found.
- [CREDENTIALS_UNSAFE] (SAFE): No API keys, tokens, or hardcoded secrets are present. The skill correctly instructs the user to manage authentication through a dedicated connection management command (RUBE_MANAGE_CONNECTIONS).
- [DATA_EXFILTRATION] (SAFE): The skill does not access sensitive local file paths or attempt to send unauthorized data to external domains.
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill relies on dynamic tool schemas fetched via RUBE_SEARCH_TOOLS. While this creates a surface where a remote server could provide instructions that influence the agent, this is an inherent property of dynamic tool-calling systems and no malicious exploitation was observed in the static file.
Audit Metadata