The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill generates executable code by ingesting user prompts and surrounding project context, creating a surface where malicious input could influence the generated output.
Ingestion points: Prototype requirements are derived from the "user's prompt" and "surrounding code" as specified in SKILL.md.
Boundary markers: The instructions do not define delimiters or specific warnings to isolate user instructions from the logic generation process.
Capability inventory: The skill allows the agent to generate and write HTML, JS, and TSX files (LOGIC.md, UI.md) and to execute project-defined task runners.
Sanitization: No specific input validation or sanitization measures are prescribed for the code generation process.
[DYNAMIC_EXECUTION]: The skill generates and facilitates the execution of new scripts at runtime based on user-provided questions.
Evidence: The skill instructions involve creating standalone HTML files with embedded logic (LOGIC.md) and React components for UI variations (UI.md).
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to run the generated prototypes within the project environment.
Evidence: SKILL.md refers to executing project task runners such as pnpm, python, and bun to launch prototypes.

prototype