The Agent Skills Directory

[SAFE]: The skill serves as a 'meta-skill' designed to help developers scaffold, evaluate, and package other skills. Its core functionality is instructional and organizational.
[SAFE]: Analysis of the SKILL.md frontmatter shows that it limits tool access to a standard set of developer CLIs (claude-code, cursor, etc.), which is consistent with its stated purpose of skill authoring.
[SAFE]: The references/ directory contains extensive documentation on safety policies, including trust-security-method.md and resource-boundaries.md, which advocate for least-privilege principles, secret scanning, and avoiding network-capable scripts without bounded policies.
[SAFE]: The agents/interface.yaml explicitly forbids remote inline execution and restricts metadata policy, demonstrating a security-conscious design for the execution environment.
[SAFE]: All referenced scripts (e.g., yao.py, governance_check.py) are described as local Python utilities for validation and report generation. No hardcoded credentials or suspicious remote execution patterns were found in the provided documentation.
[SAFE]: The skill includes a 'User Memory Policy' that explicitly forbids scanning private logs (shell history, mail, etc.) and requires human approval before applying any adaptive changes to source files.

yao-meta-skill