The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it dynamically determines which packages and components to install based on untrusted project data files.\n
Ingestion points: {PROJECT_PATH}/brief/install-manifest.md and {PROJECT_PATH}/config.json.\n
Boundary markers: None; the skill lacks delimiters or instructions to ignore embedded commands within the processed files.\n
Capability inventory: Utilizes the Bash tool to perform npm install and npx executions based on instructions found in the external manifest.\n
Sanitization: No validation or sanitization of manifest content is performed before passing the instructions to the shell for execution.\n- [COMMAND_EXECUTION]: Executes shell commands to initialize tech stacks, build projects, and install components. This includes commands like npx create-next-app, npm install, and npx shadcn, as well as build verification commands like npx next build.\n- [EXTERNAL_DOWNLOADS]: Fetches and installs software packages from the npm registry and other standard tool repositories during the scaffolding and initialization phases.

scaffold