howitworks2 baje AI se debugging jua hai: Superpowers systematic-debugging skill se root cause pehle
AI mein error paste karna, 'fixed' sunna aur naya error paana debugging nahi, jua hai. obra/superpowers (Jesse Vincent, K-9 Mail ke founder; 3 lakh+ GitHub stars, MIT license) Anthropic ke official Claude plugin marketplace mein hai. Iski systematic-debugging skill ka Iron Law hai: NO FIXES WITHOUT ROOT CAUSE INVESTIGATION FIRST. AI 4 phases mein chalta hai: investigation (error poora padho, reproduce karo, recent changes/git diff dekho), pattern (chalta hua code dhoondh kar farq nikalo), hypothesis (ek theory, ek chhota change), aur fix (pehle failing test, phir ek fix, phir verify). 3 ya zyada fix fail hon toh AI rukta hai aur architecture par aapse baat karta hai.
Links
Aaj ye karo
- Claude Code kholo aur chalao: /plugin install superpowers@claude-plugins-official
- Cursor mein Agent chat mein: /add-plugin superpowers (ya marketplace mein 'superpowers' search karo).
- Codex app mein: sidebar mein Plugins, Coding section mein Superpowers ke aage + dabao.
- Agla bug aaye toh AI ko seedha bug batao: systematic-debugging skill khud load hogi aur pehle root cause dhoondhegi.
- Agar AI 3 fix ke baad bhi atka hai, toh uski baat suno: problem architecture mein ho sakti hai.
/plugin install superpowers@claude-plugins-official/add-plugin superpowersSources
obra/superpowers README aur skills/systematic-debugging/SKILL.md, GitHub API (312,122 stars, 4 Nov 2026), Wikipedia; 4 Nov 2026 ko check kiya. Video ke terminal, error aur chat examples illustrative hain.