ui4
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
grepto scan local project files inpackages/ui/src/for icon dependencies and variable definitions. These are standard development operations scoped to the repository's source code. - [INDIRECT_PROMPT_INJECTION]: The skill identifies a data ingestion surface where it reads external design context from Figma URLs provided by the user using
mcp_figma2_get_metadataandmcp_figma2_get_design_context. While the skill processes this external data to generate CSS styles and does not explicitly define sanitization steps, this activity is the core function of the skill and no exploitable capability chains were detected. - [REMOTE_CODE_EXECUTION]: The skill instructions include running local project scripts such as
pnpm run dev v4andpnpm run test:e2e. These are standard development commands for starting a local server and verifying code changes, representing no risk beyond the local environment.
Audit Metadata