box-automation

Warn

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the user to configure a third-party MCP server endpoint (https://rube.app/mcp). This establishes a dependency on an external service that is not part of the standard set of trusted providers.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from an external source (Box), which can contain untrusted content capable of influencing the agent's behavior.
  • Ingestion points: Content is retrieved from the Box environment via tools such as BOX_DOWNLOAD_FILE, BOX_SEARCH_FOR_CONTENT, and BOX_LIST_ITEMS_IN_FOLDER as described in SKILL.md.
  • Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between user instructions and external data retrieved from Box.
  • Capability inventory: The skill possesses significant write capabilities across the Box environment, including BOX_UPLOAD_FILE, BOX_CREATE_FOLDER, BOX_DELETE_FILE, and BOX_UPDATE_COLLABORATION as listed in the Tool Reference.
  • Sanitization: There is no mention of sanitization, validation, or filtering of the content retrieved from Box before it is passed to the agent's context.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 9, 2026, 01:32 PM
Security Audit — agent-trust-hub — box-automation