The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and text-based prompt templates. It does not contain executable scripts or binary files.
[EXTERNAL_DOWNLOADS]: The skill references multiple external URLs for documentation and reference implementations. All identified URLs (developers.openai.com, github.com/openai/*, platform.openai.com) belong to a trusted organization and are documented neutrally as sources for the skill's prompting guidelines.
[PROMPT_INJECTION]: The instructions encourage high autonomy and persistence (e.g., 'finish end-to-end in one turn', 'bias to action'). However, these instructions are provided within the context of a coding assistant's primary purpose and include explicit safety constraints, such as forbidding destructive git commands like git reset --hard and git checkout -- unless specifically requested.
[COMMAND_EXECUTION]: The skill provides guidance on using standard developer tools such as rg (ripgrep) and git. It emphasizes safe usage patterns and the preference for dedicated tools over raw shell commands where possible.

codex-prompting