yo103jg

3 ms·

1.
by yo103jg 6mo ago
Honestly, a lot of them are. We ran SkillCompass across 881 ClawHub skills and abt 46% scored poorly on functional depth, meaning they tell Claude what the skill is but never what to actually do. We kept seeing the same pattern: a name, a v
2.
Show HN: SkillCompass – open-source quality evaluator for your AI skills(github.com)
2 points by yo103jg 6mo ago|0 comments
3.
Show HN: SkillCompass – Diagnose and Improve AI Agent Skills Across 6 Dimensions(github.com)
2 points by yo103jg 6mo ago|0 comments
4.
Ask HN: How do you know if a tweak to your AI skill made it better?
9 points by yo103jg 6mo ago|4 comments
5.
by yo103jg 6mo ago
If you say it’s like opening a loot box in csgo, then I totally get it. i used to be pretty good at that game, haha
6.
by yo103jg 6mo ago
Project name: SkillCompass Project description: A tool for testing, diagnosing, and improving AI agent skills. It helps make skill quality easier to inspect, compare, and improve over time instead of relying on guesswork. Would also love pe
7.
by yo103jg 6mo ago
My current split: Claude for code, Gemini for harder reasoning, ChatGPT for more structured output. ChatGPT is still useful, but mostly for tasks where formatting, organization, and response shape matter. If I’m judging mostly on raw capa
8.
by yo103jg 6mo ago
Sometimes it feels like vibe coding lowers the cost of creating new skills so much that we end up with skills for making skills, and then more skills for evaluating those skills. At some point the question stops being “how do we evaluate al