The Agent Skills Directory

[SAFE]: The skill implements a security-conscious 'fail-closed' design pattern for tool definition. It defaults security properties such as is_read_only, is_destructive, and is_concurrency_safe to False, ensuring that any tool created without explicit configuration is treated with maximum caution.
[COMMAND_EXECUTION]: The skill's primary function is to generate Python code based on user-provided schemas and logic. While this involves script generation, it follows a strict template designed to guide users toward creating modular and validated tool classes. This is an intended functionality for a developer-oriented tool builder.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file path access, or unauthorized network operations were detected. The included base class includes output truncation logic (max_result_chars) to prevent unintentional context window exhaustion from large tool outputs.
[PROMPT_INJECTION]: The instructions in SKILL.md are descriptive and technical, focusing on the interviewing process for tool definition. There are no attempts to override agent safety guidelines or hide malicious instructions in the metadata or body content.

agent-tool-builder