The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The initialization script scripts/init-artifact.sh attempts to install the pnpm package manager globally using npm install -g pnpm if it is not detected in the environment. This represents a system-wide modification and typically requires elevated permissions.
[COMMAND_EXECUTION]: The script scripts/init-artifact.sh uses the user-provided project name to update project files via sed. The interpolation pattern used ('s/.../'"$PROJECT_NAME"'.../') is vulnerable to sed injection if the project name contains the delimiter character (/), which could lead to script failure or unexpected file corruption.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a vulnerability surface by interpolating untrusted user input (the project name) into shell commands and project configuration files.
Ingestion points: Project name argument in scripts/init-artifact.sh.
Boundary markers: None are present to sanitize the input against shell or tool-specific metacharacters.
Capability inventory: File system modification, shell command execution, and package installation.
Sanitization: No validation or escaping is performed on the project name before its use in scripts.
[DYNAMIC_EXECUTION]: The setup process uses node -e to dynamically execute JavaScript code strings for the purpose of parsing and modifying tsconfig.json files. While functional, this represents a pattern of runtime code execution during the skill's setup phase.

artifacts-builder