Skill completed and first test done.
TL;DR · AI Summary
Unclear requirements make library selection meaningless; need to clarify data models first.
Key Takeaways
- Unclear requirements make library selection meaningless.
- Need to clarify data models before choosing a library.
- Test report at https://t.co/NMKfKmENRZ
Outline
Jump quickly between sections.
Completed skill writing and conducted the first test.
Discussed and researched the best WYSIWYG Markdown editor open-source libraries.
Final report located at https://t.co/NMKfKmENRZ.
Unclear requirements make library selection meaningless; need to clarify data models.
Mindmap
See how the topics connect at a glance.
查看大纲文本（无障碍 / 无 JS 友好）
- 技能测试与库选择
- 技能完成
- 调研最佳Markdown编辑库
- 结论
- 需求不明确
- 需明确数据模型
Highlights
Key sentences worth saving and sharing.
The conclusion is that my requirements were not clear enough; selecting a good data model makes library selection meaningful.
Final report here: https://t.co/NMKfKmENRZ
Writing a Skill based on this idea, using Claude Code for inference and Codex as the host, to see the results.
[Sunward Tree](https://x.com/vista8)
The Skill is done, and I've completed the first test. I researched and discussed the best open-source WYSIWYG Markdown editor libraries. The final report is here: 32kw.com/view/f4acd0c. The conclusion is that my requirements weren't clear enough; choosing the right data model makes library selection meaningful, hahaha. https://t.co/pKfwp3SZXe
Sunward Tree
@vista8
11h
Read a paper called HeavySkill, which was very interesting. It involves multiple AIs thinking independently in parallel to generate separate reasoning paths, then using another round of reasoning to synthesize all ideas into a final answer. According to the paper's test results, the quality of responses improves significantly. I'm currently writing a Skill based on this approach, with Claude Code for reasoning and Codex as the host, to see how it works. blog.qiaomu.ai/heavyskill-hea
