The Agent Skills Directory

[SAFE]: The skill instructions define a rigorous planning process without any malicious patterns. The skill incorporates mandatory user approval steps (AskUserQuestion) before any implementation begins, ensuring human-in-the-loop control. It utilizes sub-agents (file-finder, web-researcher) for scoped tasks, which is a standard and safe orchestration pattern. File access is restricted to searching for research and plan documents in specified directories, posing no risk of unauthorized data exposure. The skill also enforces test-driven development (TDD) principles and explicitly defines success criteria and risk classifications.

writing-plans