brand-identity
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a multi-agent collaboration workflow that ingests untrusted user data (e.g., brand direction, business domain) and stores it in a shared workspace file (_workspace/00_input.md). This data is subsequently consumed by various agents in the pipeline, creating a surface for indirect prompt injection attacks where malicious instructions could potentially influence the agents' outputs or the final review process.\n
- Ingestion points: User input is processed in Phase 1 and saved to
_workspace/00_input.mdinskill.md.\n - Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' prompts when handling the external data.\n
- Capability inventory: The skill workflow utilizes file system operations for writing deliverables to the
_workspace/directory andSendMessagefor inter-agent communication.\n - Sanitization: There is no evidence of sanitization, validation, or escaping of the user-provided content before it is processed by the agent team.
Audit Metadata