The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a specific multi-line bash script template for the agent to manage temporary git worktrees and branches. This script utilizes git, tar, and mktemp to isolate prototype code and commit it to a separate branch before cleaning up the temporary directory.
[DYNAMIC_EXECUTION]: The skill's primary function is the generation of runnable code, including self-contained HTML files for logic prototypes and JSX/TSX components for UI prototypes. It explicitly instructs that these artifacts should be executed using project-standard task runners such as pnpm, python, or bun.
[INDIRECT_PROMPT_INJECTION]: The agent is instructed to use the user's prompt and the surrounding codebase to determine the prototype's logic, state model, and visual structure, creating a surface for indirect instructions to influence the generated code.
Ingestion points: User-provided prompts and existing repository files (e.g., backend modules or page components) are used as context for generation.
Boundary markers: The instructions do not define explicit delimiters to separate user data from instructions when interpolating context into the prototype generation process.
Capability inventory: The skill allows filesystem writes, execution of git-based shell commands, and invocation of local task runners.
Sanitization: There is no description of sanitization or validation of the input data (prompt or code) before it is processed by the agent to create executable prototypes.

prototype