od-deck-deck-guizang-editorial
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user arguments and external design tools, which is then used to influence file-authoring tasks.
- Ingestion points: User-provided
$ARGUMENTS, content within the agent's current session context, and the instructions returned by theprepare_open_design_brieftool in theSKILL.mdfile. - Boundary markers: There are no explicit delimiters or warnings to ignore embedded instructions within the processed content or the brief template.
- Capability inventory: The skill directs the agent to use its own file-editing tools and the
register_open_design_artifacttool based on external instructions. - Sanitization: No validation or sanitization is performed on the user input or the data retrieved from the Open Design design systems before it influences file-writing actions.
Audit Metadata