box-automation
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes workflows that ingest untrusted data from Box which could contain instructions intended to manipulate agent behavior.
- Ingestion points: Data enters the agent's context through tools like BOX_SEARCH_FOR_CONTENT, BOX_LIST_ITEMS_IN_FOLDER, and BOX_GET_FILE_INFORMATION (SKILL.md).
- Boundary markers: The instructions lack guidance on using delimiters or system-level prompts to treat retrieved data as untrusted content.
- Capability inventory: The agent is granted capabilities to write, delete, and share files, increasing the potential impact of a successful injection (SKILL.md).
- Sanitization: No content validation or sanitization routines are mentioned for external data inputs.
Audit Metadata