brainstorm-experiments-new
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided product concepts via the
$ARGUMENTSplaceholder and explicitly instructs the agent to read user-provided files (e.g., market research or landing page mockups). This architecture creates a surface where malicious instructions embedded within the user input or files could attempt to manipulate the agent's behavior. - Ingestion points:
$ARGUMENTSplaceholder inSKILL.mdand user-provided files referenced in the instructions. - Boundary markers: Absent. The skill does not use delimiters or explicit instructions to ignore commands hidden within the ingested data.
- Capability inventory: The skill utilizes file-reading capabilities and file-writing capabilities (saving as markdown). It does not explicitly define network operations or shell execution in its instruction set.
- Sanitization: Absent. The untrusted input is interpolated directly into the system context without filtering or validation.
Audit Metadata