The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The scripts init-artifact.sh and bundle-artifact.sh download and install numerous packages from the NPM registry at runtime, including vite, parcel, and various @radix-ui components. While these are standard tools, the unpinned nature of some dependencies and runtime installation pose a supply chain risk.
[COMMAND_EXECUTION] (MEDIUM): The skill relies heavily on shell scripts to orchestrate the build process. It uses node -e in init-artifact.sh to dynamically execute JavaScript for patching tsconfig.json and tsconfig.app.json. This pattern of runtime code generation and execution is a security concern if the input data were to be compromised.
[PRIVILEGE_ESCALATION] (MEDIUM): The init-artifact.sh script attempts to perform a global installation of pnpm (npm install -g pnpm) if it is not detected. Modifying the global system environment is a high-impact action that typically requires elevated permissions.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill has an attack surface where untrusted data (the project name) is interpolated into shell commands.
Ingestion points: scripts/init-artifact.sh (argument $1).
Boundary markers: None present.
Capability inventory: Subprocess calls (pnpm create, pnpm install), file writes (cat > ...), and directory navigation.
Sanitization: None detected for the project name input.

web-artifacts-builder