design-frameworks
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill interacts with local development environment tools and design framework CLIs, such as Astryx and shadcn, to probe for capabilities and apply presets to the project. These operations are limited to local binaries and require user confirmation before any modifications are performed.- [EXTERNAL_DOWNLOADS]: The skill fetches design specifications and documentation from well-known platforms including GitHub, Meta/Facebook's Astryx documentation, and the official shadcn registry.- [DYNAMIC_EXECUTION]: Local test files within the skill package, specifically
tests/test_contract.py, use dynamic loading to execute the project's internal validation script,scripts/validate_fixtures.py, during development tests.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data from UI component registries and project-local configuration files, creating a potential surface for indirect instructions. - Ingestion points: Ingests project manifests (e.g.,
package.json,components.json) and external registry JSON objects. - Boundary markers: Implements strict "fit plan" and "operation receipt" boundaries to isolate external input from the agent's core instructions.
- Capability inventory: Includes the ability to write project files and execute local framework CLI commands.
- Sanitization: Enforces manual human review of all file diffs and command previews, alongside a "Never override" policy for credentials and core project logic.
Audit Metadata