The Agent Skills Directory

[SAFE]: The skill defines clear safety boundaries, specifically prohibiting production access and direct staging environment mutation. All deployment occurs through CI/CD pipelines via the GitHub CLI (gh) after explicit user review and approval of the PR draft and commit content. The skill also enforces session-specific approval for read-only server inspection, ensuring that SSH access is not persistent or assumed across sessions. Furthermore, it explicitly disables model invocation in the frontmatter, limiting the agent's capability to trigger other potentially dangerous skills.

staging-fix