The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill relies on user-provided prompts to generate functional code, creating a surface where malicious instructions in the input could influence the generated output. * Ingestion points: User prompts and existing code context are used to define the prototype's logic and user interface (SKILL.md). * Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to prevent the agent from obeying commands embedded in user-supplied data. * Capability inventory: The agent writes new files (HTML, JS, Python) and is directed to execute them via the project's task runners (SKILL.md). * Sanitization: There is no requirement for the agent to sanitize or validate user input before incorporating it into the generated prototype logic.
[DYNAMIC_EXECUTION]: The skill instructs the agent to generate executable scripts and HTML files at runtime and then run them. * Evidence: The workflow involves writing logic into script blocks in HTML files or creating separate UI components and executing them using tools like pnpm, python, or bun.
[COMMAND_EXECUTION]: The skill explicitly directs the agent to execute shell commands to run the newly created prototypes. * Evidence: SKILL.md specifies that prototypes should be Trivial to run using one command like pnpm, python, or bun.

prototype