constructive-blocks
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches component registry metadata and source code from the author's official GitHub Pages repository (
constructive-io.github.io). These downloads are part of the standard installation workflow for the UI library.- [COMMAND_EXECUTION]: Utilizes package managers (pnpm, npx, bunx) to run theshadcnCLI for project inspection, component search, and installation. These are routine development operations within the scope of the skill's purpose.- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from the component registry, including metadata and item documentation. It proactively mitigates the risk of indirect prompt injection through clear instructions to treat external data as evidence rather than authorization. - Ingestion points: Component metadata and definitions fetched from
https://constructive-io.github.io/blocks/r/via CLI tools. - Capability inventory: The skill performs file system modifications (installing components) and CLI executions via the
shadcntool. - Boundary markers: Includes explicit guidelines to maintain application-owned security boundaries and to avoid deriving authorization from registry manifests.
- Sanitization: Relies on project-level verification steps, including TypeScript typechecking, production builds, and manual review of code changes.
Audit Metadata