The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to provide standard shell commands for running prototypes, such as pnpm, python, or bun. These are used for legitimate project tasks and do not involve unauthorized execution.
[DYNAMIC_EXECUTION]: The skill generates executable artifacts, including single-file HTML logic demos and dynamic UI components. These are functional requirements of the prototyping skill and follow standard development practices.
[INDIRECT_PROMPT_INJECTION]: The skill generates artifacts that ingest untrusted user input.
Ingestion points: User interactions with buttons in LOGIC.md and URL search parameters (?variant=) in UI.md.
Boundary markers: None specified in the instructions.
Capability inventory: Execution of JavaScript logic in a standalone HTML file (LOGIC.md) and conditional component rendering within the host application (UI.md).
Sanitization: Not explicitly mentioned; the skill prioritizes rapid 'throwaway' logic over production-grade input validation. As this is for internal prototyping, the risk is negligible.
[SAFE]: No obfuscation, hardcoded credentials, sensitive data exfiltration, or privilege escalation patterns were detected. The skill follows secure prototyping practices, such as isolating logic from the DOM and gating prototype features in production builds.

prototype