skills/hachej/boring-stack/shadcn/Gen Agent Trust Hub

shadcn

Warn

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: MEDIUMDYNAMIC_CONTEXT_INJECTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [DYNAMIC_CONTEXT_INJECTION]: The file SKILL.md contains a shell command npx shadcn@latest info --json within the dynamic context syntax (!command). This command executes automatically when the skill is initialized to provide project metadata. While the command itself is used for discovery, the use of dynamic context for silent shell execution is a potential security concern.
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions for the agent to download and install code from external sources using the shadcn CLI. This includes community registries and arbitrary GitHub repositories specified by the user (e.g., owner/repo/item). Although the shadcn tool is well-known, the content added from non-official repositories is unverified. The skill mitigates this risk by instructing the agent to perform audits using --dry-run, --diff, and --view flags before finalizing installations.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external or user-provided data, specifically DESIGN.md files and documentation content fetched from remote URLs. This creates a surface for indirect prompt injection where malicious instructions embedded in a design specification or registry documentation could influence the agent's behavior during the design system creation process.
  • Ingestion points: Reads and processes the content of DESIGN.md (root directory) and documentation/example code fetched via npx shadcn@latest docs.
  • Boundary markers: No explicit delimiters or ignore-instructions are used when processing the contents of design specifications.
  • Capability inventory: The skill has access to the Bash tool for running CLI commands and script execution, as well as file system operations for editing project files.
  • Sanitization: The skill relies on the agent's interpretation of the design documents without pre-processing or validation of the input content.
  • [COMMAND_EXECUTION]: The skill frequently executes the shadcn CLI through package runners such as npx, pnpm dlx, and bunx to manage project initialization, component installation, and theme application. These operations involve structural changes to the project and modification of configuration files.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 12, 2026, 12:49 PM