kmp-compose-design-system
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill utilizes external, user-provided files such as
PRODUCT.mdanddocs/design-system.mdas context for naming components and defining token values. This ingestion surface is used for legitimate project configuration. - Ingestion points:
PRODUCT.mdanddocs/design-system.mdlocated in the project root. - Boundary markers: Absent; the files are processed as plain text design context.
- Capability inventory: The skill includes Python scripts capable of file system modification and code scaffolding within the project directory.
- Sanitization: Configuration values are extracted via regular expressions or processed directly by the agent context.
- [EXTERNAL_DOWNLOADS]: The skill references established libraries and vendor-owned resources from well-known platforms.
- Evidence: Recommends the
Pillowlibrary for image color extraction during palette generation. - Evidence: Depends on the official
detekt-apifrom Maven Central for static analysis. - Evidence: References the author's own GitHub repositories for
tailwind-composeandshadcn-composeas complementary tools.
Audit Metadata