The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill facilitates the generation and execution of logic prototypes (HTML/JS) and UI components (TSX) to answer specific design questions. It uses project-standard task runners such as pnpm, python, and bun to run these prototypes in a local development environment.- [COMMAND_EXECUTION]: The workflow involves Git operations, including creating throwaway branches, committing prototype code, and pushing to remote repositories to capture the prototype as a primary source for decision-making.- [INDIRECT_PROMPT_INJECTION]: The skill processes user prompts and existing project source code to generate prototypes. It mitigates injection risks by including a dedicated Anti-Patterns section that instructs the agent not to treat external content, logs, or generated output as trusted instructions. Evidence Chain: Ingestion points include user prompts and project code; Boundary markers are established in the Anti-Patterns and Verification Protocol sections; Capabilities include shell execution via standard task runners; Sanitization is provided via explicit logic gates in the instructions.

prototype