🚨 یک Skill برای Claude در GitHub بیشتر از ۱۶۴ هزار ستاره گرفته!
اسمش Ponytail هست؛ ابزاری که به ایجنت هوش مصنوعی کمک میکنه مثل یک برنامهنویس باتجربه فکر کنه: قبل از نوشتن کد اضافه، دنبال سادهترین راهحل ممکن بگرده.
همون برنامهنویسی که ۵۰ خط کد رو میبینه و اگر بشه، به یک خط تبدیلش میکنه!
نسخه پنجم این پروژه منتشر شده و طبق نتایج بنچمارک سازنده، خروجیهاش قابلتوجهه:
🔹 ۵۳٪ کد کمتر
🔹 ۴۱٪ زمان کمتر
🔹 ۲۶٪ هزینه کمتر
🔹 ۴۵٪ توکن کمتر
اما نکته جالبتر اینه که با وجود کاهش حجم کد، کیفیت تستها هم بهتر شده.
طبق نتایج اعلامشده، ۹۸٪ از بخشهای حساس کد با تست همراه بودن؛ درحالیکه این عدد بدون Ponytail حدود ۶۸٪ بوده.
یک مثال واقعی:
فرض کن از ایجنت میخوای یک ابزار انتخاب تاریخ بسازه.
بدون Ponytail ممکنه از صفر شروع کنه و ۳۳۵ خط کد بنویسه.
اما با Ponytail، بهجای ساختن دوباره چیزی که از قبل وجود داره، از قابلیت داخلی مر��رگر استفاده میکنه و راهحل رو در ۱۰ خط پیادهسازی میکنه.
چرا این ��وضوع مهمه؟
وقتی با AI Agentها برنامهنویسی میکنی، کد اضافه فقط مصرف توکن رو بالا نمیبره؛ بررسی، دیباگ و نگهداری پروژه رو هم سختتر میکنه.
Ponytail قبل از نوشتن کد، ایجنت رو تشویق میکنه راهحلهای موجود رو بررسی کنه، از قابلیتهای داخلی استفاده کنه و فقط چیزی رو بسازه که واقعاً لازمه.
این پروژه با ابزارهایی مثل Claude Code، Codex، Cursor و GitHub Copilot سازگاره و با مجوز MIT بهصورت رایگان منتشر شده.
📌 نکته مهم: هدفش صرفاً نوشتن کمترین تعداد خط نیست؛ هدف، کمکردن کدهای غیرضروری بدون حذف تستها، اعتبارسنجی و ملاحظات امنیتیه. درصدهای بالا هم نتایج بنچمارک سازنده هستن و تضمینی برای همه پروژهها نیستن.
🔗 لینک پروژه در GitHub:
https://t.co/yzmAsSvGwH
One developer's personal Claude Code setup just became one of the most-starred repos on GitHub.
278,000+ stars. 41,000+ forks. Free and MIT-licensed.
It's called Everything Claude Code and it's basically a full engineering team you install once:
• 71 agents for planning, code review, build repair, security and architecture
• 311 skills covering TDD, research, frontend, data, ML and docs
• Hooks and memory so Claude remembers what matters between sessions
• Continuous learning that turns repeated wins into reusable skills
• AgentShield, which scans your prompts, hooks, MCP config and agent files for secrets and risky permissions
The workflow it forces on your agent:
plan, test, implement, review, verify, remember, improve
Instead of re-explaining your process in every prompt, it becomes how Claude works by default.
It works best with Claude Code, and it also has adapters for Codex, Cursor, Gemini, OpenCode and more.
Setup is one command.
Repo below.
https://t.co/1PbzeenMCe
THIS IS A PURE F*CKING GOLDMINE
Anthropic just shared one of the best tricks for Claude Code.
The idea: let Opus 5.5 write the plan as a script, and let a swarm of cheaper agents run it.
Bigger jobs and way fewer Opus tokens, because the script holds the plan and Opus only sees the final answer.
You just launch Claude Code like this:
claude --model opus --effort ultracode
here's the split:
→ Opus 5.5 turns your task into a workflow script
→ Haiku 5.5 agents scan every file in parallel
→ Sonnet 5.5 agents patch each fix in its own isolated copy
→ you stop burning Opus context on a hundred agent transcripts
Opus only shows up at three moments:
→ writing the workflow before the run
→ reading the verified results after it
→ signing off before anything ships
Opus plans. Haiku scans. Sonnet ships.
JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split
the full tree
> Opus 5.5 runs the main session and writes the workflow
> scan audits every file on Haiku 5.5
> edit patches each fix on Sonnet 5.5 in an isolated copy
> verify cross-checks every finding on Sonnet 5.5
> Opus 5.5 reads only what survived verification
> JEV routes micro-forks in milliseconds
paste the tree and this prompt into Claude Code ↓
"Rebuild my Claude Code setup around this tree:
1. Run the main session on Opus 5.5 with ultracode on.
2. In every workflow, run read-only scan stages on Haiku 5.5 and edit stages on Sonnet 5.5, each edit in its own isolated copy.
3. Add one rule to CLAUDE.md: every finding gets an independent verifier agent before it's reported, and fix loops stop after two rounds with no progress.
4. Route fast deterministic forks (tool choice, path selection, retries) through JEV.
Show me every change as a diff first. No edits until I say go."
↳ https://t.co/1MiKDqGANA
save it, it'll come in handy.
바이브코딩 하시는 분들!!
🔥 움직이는 UI 200개, 전부 100% 무료로 풀었습니다. 🔥
--> Lucian UX/UI Prompt입니다.
(제가 만든 오픈소스입니다!)
누르면 종이비행기가 날아가는 버튼, 낮과 밤이 바뀌는 스위치, 움직이는 카드와 차트 등등..
웹사이트에서 직접 눌러보고 마음에 드는 걸 고르시면 됩니다.
사용법도 간단합니다!
원하는 UI 고르기 --> "프롬프트 복사" → 클로드코드나 코덱스에 ��여넣기
(그냥 마음에 드시는 UI 결정 후 프롬프트 복붙하면 끝!)
💡 라이트와 🌑 다크 모드도 둘 다 지원합니다.
그리고 클로드코드, 코덱스에서 알아서 쓰라고 맡길 수도 있는 스킬도 준비했습니다.
설치해두고 이렇게 시키시면 됩니다.
--> "lucian-uxui 스킬로 내 홈페이지에 요금제 토글이랑 FAQ 넣어줘."
그러면 AI가 내 프로젝트 구조와 테마를 먼저 확인하고, 갤러리에서 필요한 UI를 찾아 실제 코드를 가져와 적용합니다.
스킬 설치는 터미널에서 아래 명령으로 할 수 있습니다.
npx skills add Seokwoooo/lucian-uxui
개인 프로젝트든 상업용 서비스든 무료로 쓰실 수 있습니다.
필요하신 분들 자유롭게 가져다 쓰세요 :)
https://t.co/9PpNDpBlT2
REA is most dangerous thing on the internet right now...
It lets your AI agent reverse engineer any app, down to the binary
If you use it correctly, you can rebuild any feature you pay for, for FREE
In this article, I show you how: https://t.co/3P1FNXiXzl
Your Claude loops can burn money while nothing changes
Use Jev to filter tasks and cut decision costs
Fix the trigger before another overnight run
The full Loop engineering + Jev engineering setup below ↓ https://t.co/gXjyOq2Ga1