The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill focuses on engineering best practices, such as research, validation, and understanding underlying mechanisms before implementing fixes.
[PROMPT_INJECTION]: The skill uses strong instructional language (e.g., "HARD RULE", "non-negotiable") to enforce its planning methodology. These are operational constraints designed to ensure a structured workflow and do not attempt to bypass safety filters, extract system prompts, or override core safety guidelines.
[EXTERNAL_DOWNLOADS]: The skill references standard development and testing tools such as Vitest and Playwright in the context of debugging practices. It does not contain instructions to download or execute untrusted scripts or packages from unknown sources.
[COMMAND_EXECUTION]: While the skill mentions tools like Bash and Edit, it does so by instructing the agent on the sequence of operations (e.g., do not call Bash before EnterPlanMode). It does not provide arbitrary or dangerous command strings to be executed.

plan