shadcn
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and process data from external, potentially untrusted sources, creating an attack surface for indirect prompt injection.
- Ingestion points: The agent is instructed to fetch documentation and usage examples from external URLs and to add components directly from community registries or public GitHub repositories (SKILL.md, registry.md).
- Boundary markers: The instructions explicitly tell the agent to "read the added files and verify they are correct" after adding a component to the project (SKILL.md).
- Capability inventory: The skill has shell execution capabilities via the
shadcnCLI, allowing it to write files to the project (SKILL.md). - Sanitization: No programmatic sanitization is defined; the skill relies on the agent's manual review of the imported content.
- [DYNAMIC_CONTEXT_INJECTION]: The skill utilizes dynamic context injection to provide the agent with project-specific information during initialization.
- Evidence: The
SKILL.mdfile contains the pattern!npx shadcn@latest info --json``, which executes at load time to populate the agent's context. - Context: This is a benign use of the feature, intended to retrieve essential project configuration details (such as the framework, Tailwind version, and installed components) to ensure the skill functions correctly in the user's environment.
- [COMMAND_EXECUTION]: The skill's primary purpose involves the execution of shell commands through the
shadcnCLI. - Evidence: The instructions frequently direct the agent to use package runners (npx, pnpm dlx, bunx) to execute the
shadcncommand for project management tasks. - Context: This is the core functionality of the skill. The instructions include safety guidelines for choosing the correct package runner and require explicit user approval for destructive actions like overwriting existing files.
- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of source code and configuration files from external registries.
- Evidence: The
addcommand is used to download component source files from official, third-party, and GitHub-based registries (registry.md, cli.md). - Context: This is a fundamental feature of the
shadcn/uiframework. The skill emphasizes using the CLI for these downloads rather than manual fetching, providing a structured way to integrate external code into the project.
Audit Metadata