M. Burak Akgün

@ersinkoc Abi güzel olmuş , Acaba cloudpanel benzeri node projeleri, veya Python siteler içinde destek gelir mi ? Bun ve uv runner environmentler ile ilgi çekici olabilir

github.com/uwaserver/uwas uwaserver.com

Keeping it simple: this post is now my build log. NPM packages, projects, developer tools things you can actually put to work. Let's go.

@cursor_ai Berlin :) ?

Cursor now shows you demos, not diffs. Agents can use the software they build and send you videos of their work.

Since Clowdbot showed up, the request feed has turned into a freakshow. One requested “helpfully” pastes his sk-... credential, another shoves a non-existent JWT into a UUID field and hits send like it’s totally normal. n8nbuilder.dev/templates/exam… Even though integration is simple :)

@lmstudio Yeah finally parallel executions ♥️

Introducing LM Studio 0.4.0 🎉✨ It's the next generation of LM Studio. 🪄 Deploy on servers, in CI, or anywhere 🚄 Parallel requests for high throughput use cases 🔨 New stateful REST API: use local MCPs 🎨 Complete UI revamp See what's new in this release👇🧵

@lmstudio @Zai_org Any plan for parallel executions? Still queue mode ?

GLM-4.7-Flash by @Zai_org is now in LM Studio 30B local coding agent on your Mac Available in MLX for Apple Silicon lmstudio.ai/models/zai-org…

A new #n8n #workflow #generator model that is more powerful yet runs faster than the previous model(compared to qwen2.5 coder n8n-workflow-generator) 🤗 The second in the series is live and can be downloaded on Hugging Face in BF16 and MLX 4-bit. huggingface.co/mbakgun/qwen3-…

🚀 Introducing Qwen3-Coder-30B-A3B-n8n-Workflow-Generator! 30B MoE model for n8n workflows. Less params active per token = faster inference while maintaining quality. 🔗 huggingface.co/mbakgun/qwen3-… #n8n #AI #WorkflowAutomation #MoE

@AdinaYakup @huggingface @minimax_ai Quite impressive moe skills

MiniMax-M2.1 is now live on @huggingface 🔥 huggingface.co/MiniMaxAI/Mini… ✨ 229B total / 10B active - Modified MIT license ✨ Agent-native design ✨ Outperforms Claude Sonnet 4.5 in multilingual coding ✨ Full-stack app building kudos to @minimax_ai 👏

@AlicanKiraz0 2026 yılbaşı dilekleri (ML engineer edition) 🙏 🌟 Loss'un düşük olsun, 🌟 Gradient'ların sakin kalsın, 🌟 Memory'n sabit dursun, 🌟 Model'in hızlıca converge olsun. Ne explosion, ne vanishing - > bam bam bam

Wow 🔥 mlx’de sıkı bir konfigürasyon sonrası M2.1 8bit-gs32, M3 Ultra 512 ile 185.10s/it hızına ulaştı 🦾

@raxetul Est , teşekkür ederim :)

Hoca sen ne yaptın ya? 👏

⚡ Tired of manually wiring n8n lead gen workflows? With n8nbuilder.dev you describe the funnel once and get a production-ready workflow you can import in seconds. Perfect for agencies and SaaS teams. 🚀 #n8n #leadgeneration #SaaS #automation youtube.com/watch?v=koMhJo…

Special thanks to @cerenekeye ♥️ & @AlicanKiraz0 👑

🔗 Model: huggingface.co/mbakgun/Qwen2.… 🧠 Dataset: huggingface.co/datasets/mbakg… ⚡️ App: n8nbuilder.dev Perfect for generating complex n8n workflows from natural language prompts 💯 Wishing everyone a wonderful 2026 in advance 🎉 #n8n #AI #Automation #LLM #OpenSource

New Model 🚀 Introducing: Qwen-Coder-14B-n8n-Workflow-Generator released! I fine-tuned Qwen 2.5 Coder (14B) on +2.5K real #n8n workflows using H100 GPUs. The result? An AI that speaks "n8n JSON" fluently. ⚡️🎉 Context Size: 8192

🚀 New: ready-made n8n workflow templates at n8nbuilder.dev — with real prompts, generated workflows, and one-click copy for email, CRM, social, AI and more. Explore them here: n8nbuilder.dev/templates #n8n #automation #nocode 🤖

Docker’s MCP Gateway attacks the other half: instead of hardcoding MCP servers per client, agents can use dynamic MCPs(mcp‑find, mcp‑add, code-mode) to discover and attach trusted servers on the fly(with isolation) with fewer tokens wasted on unused tools. docker.com/blog/dynamic-m…

Cloudflare’s “Code Mode: the better way to use MCP” nails the core issue: LLMs are better at writing code that calls tools than juggling them step‑by‑step, and bindings hide auth + keys so you don’t leak secrets. Worth a read: blog.cloudflare.com/code-mode/

MCP is powerful, but today’s reality is still “token & tool hell”: hardcoded configs per agent, duplicated flows, and context windows clogged with tool schemas you barely use. Agents feel more like fragile wiring diagrams than autonomous systems.