The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/init-artifact.sh script is vulnerable to command injection via the $PROJECT_NAME variable. The variable is derived from the first command-line argument and used directly in a sed command: $SED_INPLACE 's/<title>.*<\/title>/<title>'"$PROJECT_NAME"'<\/title>/' index.html. If the input contains shell metacharacters or sed delimiters, it can result in arbitrary code execution.\n- [EXTERNAL_DOWNLOADS]: The skill's scripts perform extensive runtime installations from the NPM registry. This includes installing bundlers (Parcel), frontend frameworks (Vite, Tailwind CSS), and over 40 UI component libraries. It also includes a global installation of pnpm if the tool is not found on the system.\n- [COMMAND_EXECUTION]: The initialization process involves dynamic execution of JavaScript via node -e to programmatically update tsconfig.json and tsconfig.app.json with path aliases. This introduces complexity and potential risks associated with runtime configuration generation.\n- [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection due to the lack of sanitization on the project name argument in scripts/init-artifact.sh.\n
Ingestion points: The project name is accepted as an unsanitized command-line argument.\n
Boundary markers: None. The script does not utilize delimiters or instructions to ignore embedded commands in the input string.\n
Capability inventory: The skill has the ability to execute shell commands, modify local project files, and fetch packages from external registries.\n
Sanitization: No validation or escaping is performed on the project name before its use in shell-mediated operations.

web-artifacts-builder