🇮🇷 رباتِ خَسته
876 posts

🇮🇷 رباتِ خَسته
@HazratCrypto
سکوتِ اجباری
Iran pahlavi 🇮🇷 Присоединился Ağustos 2024
671 Подписки88 Подписчики

213.000 estrellas en GitHub.
No es un modelo. No es una app. No es un framework.
Es el sistema que hace que tu agente de IA trabaje mejor.
Se llama ECC y empaqueta en un solo repo todo lo que necesitas para que Claude Code, Codex, Cursor u OpenCode funcionen como los usan los mejores.
262 skills públicas.
Skills, instincts, memoria, seguridad y flujos de trabajo listos para instalar.
Lo construyó Affaan Mustafa en el hackathon de Anthropic x Cerebral Valley.
Con Claude Code.
Y la release 2.0.0 la automatizó ella misma.
El bot de Discord, los anuncios y el deploy se desplegaron solos de principio a fin.
Lo que incluye ECC 2.0:
✅ 262 skills para coding, investigación, seguridad, media, enterprise y flujos de agentes
✅ AgentShield: escanea tu configuración de Claude Code buscando vulnerabilidades, inyecciones y secretos expuestos
✅ Modo red team/blue team con 3 agentes Opus en paralelo - atacante, defensor y auditor
✅ 1.282 tests. 98% de cobertura. 102 reglas de análisis estático.
✅ Memoria compartida entre agentes con contexto comprimido entre sesiones
✅ Compatible con Claude Code, Codex, Cursor, OpenCode, Gemini, Kiro y Zed
✅ Las mismas skills funcionan en todos los entornos sin reescribir nada
✅ 14 releases. ECC 2.0.0 hace 2 días. Activo hace 49 minutos. MIT.
El objetivo en una línea:
Dejar de reescribir los mismos prompts y convertir el comportamiento que ya funciona en infraestructura portable.
el enlace 👇
Español

Qwopus 3.6 27b-Coder is now live!
Scores a 67% on a full run of SWE bench verified with thinking completely disabled! Q5_K_M
This model is lightning fast for dense class! With a natively finetuned MTP head, it achieves 100 tps on a single 5090! The biggest upgrade here, though, is its stability in programming and tool calling within @NousResearch Hermes agent, with thinking off!
Wall time is crazy fast this way, which makes Hermes feel "native" and snappy, like they were meant for each other. The freedom of running without thinking at all makes you part of the thinking process, and you never get caught waiting 15 minutes for it to finish a thought string, like with the base models.
Thinking on and temp high, .9-1 seems to produce really incredible design and svg results. I reran the Boat survival prompt through a few turns, thinking on, and it seemed to render more fancy models in HTML canvas, but it was much more of a start-a-prompt and wait experience vs the snappy and active iteration with it disabled. It may be worth turning it off and on throughout the build process if you want to get really creative with design.
Really looking forward to seeing how this one performs for y'all! Please post comments with your opinions and use cases below! As always with our fine-tunes, mess with the temperature setting, and run them much hotter than the base!
Please check out the Boat Survival game I posted yesterday, made in 12 turns using Hermes and this model, with thinking off. Link below!
Full swe bench repo-specific breakdown also posted in the comments for those interested!
Happy building, everyone! We're looking forward to your thoughts! Quants uploading now!
huggingface.co/Jackrong/Qwopu…
English

@haghiri_codes و 70 درصد کارایی که v4 pro میکنه qwen 3.6 27b dense به رایگان لوکالی تو سیستمتون انجام میده حالا اگر یکم باهوش باشید به ۱۵۰ درصد هم میرسه😍
فارسی

9 GITHUB REPOS THAT MAKE HERMES AGENT ACTUALLY POWERFUL
most people install the base agent and wonder why it feels limited.
these are the repos experienced operators actually use.
1/ NousResearch/hermes-agent
the core of the entire ecosystem.
160k+ stars, support for 20 messaging platforms, 6 execution backends, and a built-in self-improving skill loop.
everything else on this list plugs into it.
github.com/NousResearch/h…
2/ wondelai/skills
380+ pre-built skills ready to install.
compatible with Hermes, Claude Code, and other agents.
the fastest way to give your agent real capabilities without building everything yourself.
github.com/wondelai/skills
3/ outsourc-e/hermes-workspace
the complete operator dashboard.
includes chat, memory browser, skill marketplace with 2,000+ skills, MCP manager, kanban board, and multi-agent controls.
github.com/outsourc-e/her…
4/ AMAP-ML/SkillClaw
automatically improves and organizes your skill library using real session data.
adds an extra skill evolution layer on top of Hermes’ built-in skill creation system.
705 stars.
github.com/AMAP-ML/SkillC…
5/ ZeroPointRepo/awesome-hermes-skills
a curated collection of built-in and community-created skills.
every skill is tagged by status so you can quickly find what’s useful.
great for discovering existing solutions before building your own.
github.com/ZeroPointRepo/…
6/ SamurAIGPT/awesome-hermes-agent
the operator’s reference guide.
includes tools, integrations, deployment templates, and multi-agent stack recommendations.
updated regularly.
github.com/SamurAIGPT/awe…
7/ ksimback/hermes-ecosystem
a map of the wider Hermes ecosystem.
80+ quality-filtered repositories across 12 categories.
includes live star counts and security-reviewed projects.
github.com/ksimback/herme…
8/ 0xNyk/awesome-hermes-agent
a more opinionated curated list.
focused on deployment hardening, multi-agent execution, and migration paths from OpenClaw.
github.com/0xNyk/awesome-…
9/ hermes-agent-docker
a production-ready Docker deployment for Hermes.
includes isolated gateway and sandbox environments, rootless setup, and built-in secrets management.
github.com/NousResearch/h…
nine repos.
all free.
all open source.
the base agent is just the starting point.
this is the stack people actually run.
bookmark this before your next deploy.
MIKE@mikenevermiss
English

@bookunt میشه چند مدل لوکال بهش داد که وقتی خواست بعدی رو استفاده کنه رم گرافیک رو خالی کنه بعدی رو ران کنه یا باید دستی کرد؟
فارسی

@bookunt راستی من اکانتم عوض کردم اینقدر زرنگه شناسایی میکنه بعد چند تا سوال ادامه محدودیتمو نشون میده بعضی وقتا هم زبونی میگه محدودی منم میگم نه نیستم اشتباه میکنی تا یجا میره بعد دیگه ارور قرمزو میده میگه برو پیکارت
خیلی کیووووت و باهوشه
فارسی

از روزی که Qwen max رایگانو پیدا کردم تقریبا هیچ مشکلی توی زندگیم ندارم جز اینکه دارم نمیخوابم

Emma@limunadam
از روزیکه Claude max خریدم تقریبا هیچ مشکلی توی زندگیم ندارم جز اینکه دارم نمیخوابم
فارسی

@bookunt خخخخ اره هم مکس و هم پلاس 😂😂😂
فک کنم کمتر کسیاینطوری از qwen کار بکشه حالا در همین حال لوکالش هم مثل اسب میزنم کار کنه qwen 3.6 35b a3b به همراهی ۲۷b و gemma 31b qat به صورت مرحله ای و اتو تعویضی
سرویسشون کردم 😅
فارسی

@Saeedrss @SepehrHashemi و اما پادشاه qwen codet next برای برنامه نویسی و یک مدل انلاین بالاسرش مثل z.ai پلن کدینگ یا میمو یا خود qwen 3.7 max به شدت کارو ارزون در میاره
اگرم پولدارید ک کلاد 😅
فارسی

@SepehrHashemi من با اینا کار کردم از این راضی تر بودم qwen3.6:35b-a3b-q8_0 نسخه جدیدش سریع تر بود ولی خیلی بیراهه میرفت qwen3.6:35b-a3b-mtp-q8_0 کلا در انتها نظرم عوض شد و دارم یه جور دیگه ستاپ میکنم

فارسی

@F_Mohajerani واقعا نظر دادن راجب قیافه دیگران کار درستی نیست ولی..
فارسی

@Rezqology سوال: بله نسبت به روبیکا یا ایتا یا سروش پلاس چه افزوده و ویژگی بیشتری داره؟
فارسی
















