The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes several shell commands including pnpm install, pnpm build, and npx tsc to manage dependencies and validate generated prototype projects. It also runs internal utility scripts using node during the generation workflow.
[EXTERNAL_DOWNLOADS]: The skill automatically installs additional agent skills from the trusted Vercel Labs repository (vercel-labs/agent-skills) and includes resources from well-known services such as Google Fonts and the Tailwind CSS and Lucide Icon CDNs.
[DYNAMIC_EXECUTION]: At runtime, the skill scaffolds complete project structures (React/TypeScript or HTML/Tailwind), generates source code based on extracted PRD content, and then performs compilation and build steps to ensure viability.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted markdown PRD files to generate code and UI components. It lacks specific delimiters or sanitization logic to prevent malicious instructions within the PRD from influencing the generation process.
Ingestion points: Reads files from prd/main_prd.md and prd/modules/ to extract project names, roles, and functional requirements.
Boundary markers: No specific delimiters are employed to separate external PRD content from the agent's internal instructions.
Capability inventory: Possesses the ability to write files to the local file system and execute shell commands for building and installing packages.
Sanitization: Extracted PRD text is directly interpolated into generated code and UI tooltips without escaping or validation.

prd-prototype