kernel-design
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
subprocess.runwithinscripts/fetch_ui_tokens.pyto execute a companion validation script (scripts/validate_tokens.py). This is used for internal workflow integrity and employs list-style arguments to prevent shell injection. - [EXTERNAL_DOWNLOADS]: The skill fetches design tokens in JSON format from the vendor's own GitHub repository (
ponti-studios/ui). This is a documented part of the skill's synchronization workflow between the design system and the local environment. - [DYNAMIC_EXECUTION]: The agent is instructed to run various local Python and Node.js scripts (e.g.,
scripts/contrast.py,scripts/measure_render.mjs) to perform accessibility audits and visual measurements. These tools are provided within the skill package and operate on local data. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as UI code, screenshots, and remote tokens, which represents a potential attack surface for indirect prompt injection if those sources contain adversarial instructions.
Audit Metadata