dropbox-sign-automation
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it dynamically retrieves and processes tool definitions and execution plans from an external source.
- Ingestion points: Dynamic tool schemas and recommended execution plans are ingested from the
RUBE_SEARCH_TOOLSresponse. - Boundary markers: There are no boundary markers or instructions present to prevent the agent from following malicious instructions potentially embedded in the tool metadata.
- Capability inventory: The agent has access to
RUBE_MULTI_EXECUTE_TOOLandRUBE_REMOTE_WORKBENCH, which can execute a variety of operations based on the ingested schemas. - Sanitization: The instructions do not define any validation or sanitization steps for the data returned from the remote server.
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to configure a remote MCP server endpoint at
https://rube.app/mcpto provide tool functionality. - [COMMAND_EXECUTION]: The workflow relies on dynamic command execution where tool slugs and input arguments are determined at runtime from the remote discovery service rather than being statically defined.
Audit Metadata