design-md-specification
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data from design token packs and Figma exports, which creates a surface for instructions embedded in metadata or comments to influence agent behavior.
- Ingestion points: Reads and processes content from
DESIGN.md, externally sourced design token packs, and Figma exports. - Boundary markers: The instructions do not define explicit delimiters or boundary markers for the external content when it is processed by the agent.
- Capability inventory: The skill utilizes powerful tools including
read_file,write_file,edit_file,create_file,search_code, andrun_build. - Sanitization: The skill mitigates risks by requiring strict schema validation of imported token sets against
contracts/schemas/design-tokens-spec.jsonand explicitly warning to treat token content as data rather than instructions.
Audit Metadata