The current way I check skills is to ask the AI to use “Skill authoring best practices” as the basis for checking
https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices
Since OpenAI also recently published a page about Agent Skills also used that in a latter reply.
https://developers.openai.com/codex/skills
The linked conversation below provides useful information for the skill cheeky-razor. The take away from all of this is that the AI companies often have such pages or documentation and they should be used.
It also appears that OpenAI may be working on scripts and/or skills to validate skills so check the GitHub OpenAI Skills repository often.
https://github.com/openai/skills
https://github.com/openai/skills/blob/main/skills/.system/skill-creator/SKILL.md
https://github.com/openai/skills/blob/main/skills/.system/skill-creator/scripts/quick_validate.py
Here is a link to a Claude conversation for such:
I don’t plan to delete the link in the next week but would suggest grabbing any useful information sooner rather than later.
Note: The linked conversation was the first time I worked with a skill that should work with both Claude and ChatGPT so learned a few things.
Would be interested in what you find of value from the conversation and what you learned being new to skills.
While the AI did not seem to have a concern with the quantity of meta-data lines in the frontmatter, my personal concern is that a present I have 48 unique skills, and would not be surprised if it grows to more than a few hundred, so the size of the frontmatter, while small, when having a large multiplier factor can become a concern for me.
The following is a capture from the OUTPT tab of a VS Code terminal
2025-12-21T22:32:44.965Z [DEBUG] Loaded 48 unique skills (managed: 0, user: 1, project: 41, legacy commands: 7)