token-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external data including design system documentation, token files, and theme code.
- Ingestion points: SKILL.md (Establish the source of truth) instructs the agent to read external files such as
design-system docs,token files,generated outputs,theme code, andFigma variable mapping. - Boundary markers: No explicit delimiters or instructions to ignore embedded instructions in data are provided.
- Capability inventory: The skill contains no file-writing, network communication, or command execution capabilities; it is purely instructional and analytical.
- Sanitization: No sanitization logic for ingested data is present.
- [SAFE]: The external repository URLs provided in the metadata belong to the skill's author and organization, representing standard source documentation.
Audit Metadata