The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/init-artifact.sh script performs multiple system operations including file extraction with tar and environment setup. It also attempts to install pnpm globally via npm install -g pnpm if it is not present.
[COMMAND_EXECUTION]: The scripts/init-artifact.sh script uses sed to update the <title> tag in index.html using the user-supplied project name. This input is not sanitized, which can lead to shell command failure or the injection of unintended content into the generated HTML files.
[EXTERNAL_DOWNLOADS]: The skill installs a large number of packages from the NPM registry, such as vite, tailwindcss, parcel, and various @radix-ui libraries. These are established tools in the web development ecosystem.
[REMOTE_CODE_EXECUTION]: The project initialization relies on pnpm create vite, which fetches and executes scaffolding code from the NPM registry.
[COMMAND_EXECUTION]: The initialization script uses node -e to dynamically modify tsconfig.json and tsconfig.app.json by executing inline JavaScript code.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the project name parameter.
Ingestion points: The project name enters the system as a command-line argument in scripts/init-artifact.sh.
Boundary markers: No boundary markers or 'ignore' instructions are used to separate the project name from the shell command logic.
Capability inventory: The script has the ability to run shell commands (sed, pnpm, tar), modify files, and install global packages.
Sanitization: No validation or sanitization is performed on the project name before it is interpolated into shell commands.

web-artifacts-builder