launch-shadcn-registry
Pass
Audited by Gen Agent Trust Hub on Jul 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands using
ghandgitto fork repositories, clone source code, and create pull requests. These high-privilege operations are part of the core functionality and are performed after user approval. - [EXTERNAL_DOWNLOADS]: Fetches directory configuration and project files from the official
shadcn-ui/uirepository on GitHub and performs connectivity checks on user-provided registry endpoints usingcurl. - [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests and processes metadata (names, descriptions, and features) from external
registry.jsonfiles which are subsequently used in the generation of GitHub PR bodies and social media drafts. - Ingestion points: Registry metadata is downloaded from external URLs via
scripts/validate-registry.shand parsed by the agent. - Boundary markers: None identified. External metadata fields are interpolated directly into Markdown templates for PRs and posts.
- Capability inventory: The agent has access to
gh(PR management/repository access),git(local and remote filesystem operations), andcurl(network access). - Sanitization: There is no evidence of sanitization or content validation on the description fields retrieved from external sources before they are used in generated output.
Audit Metadata