The Agent Skills Directory

[SAFE]: The skill serves as a reference guide for developer tooling. It explains how to use the doc(), spec(), and hidden utilities to format API contracts for LLM consumption. Nested within the provided examples are best practices for maintaining clean and secure agent prompts.
[DATA_EXPOSURE_PREVENTION]: The documentation explicitly promotes security best practices by demonstrating how to use the hidden marker and Annotated types to ensure sensitive information (such as API keys or internal caches) is excluded from the rendered output that the LLM sees.
[COMMAND_EXECUTION]: While the library uses Python's inspect.getsource and ast modules to parse class definitions at runtime, this behavior is standard for documentation generators and is used here to extract field descriptions rather than to execute untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill describes a system for constructing prompt context from live Python objects. It provides controls such as max_length, max_string, and max_depth to prevent data overflow and unintended prompt injection that could occur if raw, untrusted data were interpolated into the agent's context without constraints.

nooa-agentdoc