The Agent Skills Directory

[SAFE]: The skill is designed for project governance and agent lifecycle management. It includes mandatory security auditing steps (Step 5) to identify over-privileged or dangerous command configurations in new agents.
[SAFE]: The skill operates entirely on local project files such as AGENTS.md and internal directory structures. It lacks any network-facing capabilities or external script execution patterns.
[SAFE]: Documentation in references/admission-guide.md establishes clear security boundaries and criteria for agent admission, explicitly prohibiting agents that lack safety guardrails or guide users toward high-risk operations.

agent-creator