Nurhayat Yurtaslan (@nurhayatdev) header
Nurhayat Yurtaslan (@nurhayatdev) profile picture

Software Engineer | Exploring AI agents, open-weight models & visual systems. Daily AI updates, architecture & prompts. La imaginación › el conocimiento.

Turkey
Joined August 2022
🧾 Month-end close is getting a new workflow in ChatGPT. @OpenAI’s Accounting plugin can turn invoices and receipts into draft entries, gather records from email, files and connected accounting apps, then help reconcile accounts and prepare reports. It flags unclear items and possible duplicates for human review—so people stay in control of the books. It connects with QuickBooks, Xero, BILL, FreshBooks and NetSuite. Enterprise workspaces require admin approval. Details: openai.com/business/solution…
3
⚡ Compose code audits, now in the Android CLI. @AndroidDev’s new performance skill scans any PR, file or screen across seven areas. The updated adaptive skill adds foldable tabletop layouts and desktop shortcuts, tooltips and custom cursors. Official details: developer.android.com/blog/p…
8
Still free — only inside the @opencode client. `/models` → Hy4 flash Free (`opencode/hy4-flash-free`). Docs: $0 in/out, zero retention. Other clients get HTTP 403. They said 10 more days on 18 Nov. Try @TencentHunyuan's model in the CLI first? opencode.ai/docs/console/mod…
6
🛡️ @TheHackersNews: PHP security updates are due today. PHP schedules fixes for all supported versions at 14:00–14:30 UTC. Critical issues need non-default settings or uncommon setups; fixed versions will come with the release. Be ready to patch: php.net/archive/2026.php
5
🚨 @Ubiquiti: UniFi Access Bulletin 073 fixes 7 flaws in 4.3.22 and earlier. Update to 4.3.24+. CVE-2026-132802 (CVSS 9.9) allows root code execution by a low-privilege network attacker. Six more bugs include SSRF and authorization issues. Patch: community.ui.com/releases/Se…
4
Exploring a new visual concept with @canva. ✦ The Internet of Minds ✦ Created with Canva AI ✦ 1080 × 1350 Prompt in the replies ↓
1
3
Create a vertical 1080x1350 px retro-scientific editorial poster for X. Headline: "THE INTERNET OF MINDS". Illustrate interconnected human brains as an intricate communication network, combining vintage scientific diagrams with futuristic computing elements. Use forest green, aged ivory and black. Fine engraving details, archival paper textures and elegant serif typography. Inspired by vintage scientific journals reimagined through modern graphic design.
2
🎬 Claude is gaining a creative studio. @higgsfield’s MCP 2.0 brings pro tool integrations, multi-agent creative workflows and built-in quality control to Claude Fable 5.2—letting creators move from brief to production in one conversation. See the demo: teapawt.net/higgsfield/status/2123…
Introducing Higgsfield MCP 2.0 for Claude. Your entire creative studio, now inside Claude Fable 5.2. > Pro-level tool integrations > Multi-agent creative workflows > Built-in quality control Now available in Claude. Try it free for 3 days with unlimited generations in MCP 2.0
3
Gemini 4 Neon is now generally available in GitHub Copilot for agentic coding, multi-step tool use and knowledge work. GitHub says it brings near-Carbon quality at lower token cost per task, faster responses and fewer unwanted code edits. ⚡ It’s rolling out to Pro+, Max, Business and Enterprise across VS Code, Visual Studio, Copilot CLI, the coding agent, GitHub.com, mobile, JetBrains, Xcode and Eclipse. Admins can manage Business/Enterprise access in model policy. @github github.blog/changelog/2026-1…
2
32% lower cost per task than Carbon, at default high thinking. Gemini 4 Neon (`gemini-4-neon`) is on the paid Gemini API: 1M context, $3 / $15 per 1M tokens. @GoogleDeepMind places it between Hydrogen and Carbon. Will you switch any workloads over? blog.google/innovation-and-a…
4
Gemini 4 Neon is rolling out to Google AI Pro and Ultra subscribers on mobile, desktop and web. Google says it pairs frontier-level reasoning with fast responses. ⚡ The gemini-4-neon model ID is also listed in Google Cloud’s model catalog. Availability is subscriber-gated at launch. @GeminiApp cloud.google.com/gemini-ente… teapawt.net/GeminiApp/status/21231…
Starting today, Gemini 4 Neon is available to Google AI Pro and Ultra subscribers on mobile, desktop, and web. This new model excels at work that requires frontier-level reasoning and fast responses. 🧵

4
TeamCity’s Perforce settings could open a path to remote code execution. ⚠️ JetBrains rates CVE-2026-132231 Critical. It affects TeamCity On-Premises before 2025.11.7 or 2026.1.3, depending on the release branch. The CVSS vector requires low privileges—this is not an unauthenticated entry point. Patch to the fixed release for your branch. Check the TeamCity build and JetBrains’ security fixes: jetbrains.com/privacy-securi… @jetbrains
7
A product agent that keeps working after you merge. @GoogleLabs launched Stitch Loop today: it watches your codebase and metrics, then opens a GitHub PR from a cloud sandbox. Nothing merges until you approve. Would you start it on the security loop? blog.google/innovation-and-a…
5
Codex can still read files that hold your API keys. @OpenAIDevs on the Codex sandbox: default :workspace allows broad filesystem reads, keeps `.git` read-only, and turns network off. Package installs need a network grant. Changed the default profile yet? openai.com/index/inside-the-…
8
220B parameters. Only ~12B run per token. @TencentHunyuan released Hy4 flash: 1M context, text + image. Official vLLM image: vllm/vllm-openai:hy4-flash. Check the Tencent Hy License before commercial use. Would you self-host 1M context at ~12B active? huggingface.co/tencent/Hy4-f…
6
Python joins Vercel Sandbox at 1.0. 🐍 The Python SDK now has a stable API and feature parity with the TypeScript SDK. It supports isolated code runs, AnyIO for asyncio/Trio, and a synchronous interface—useful for running AI-generated code or tests from Python apps. Install: `uv add 'vercel-sandbox>=1.0'`. Upgrading from 0.x? Check the release notes: command input and output now use bytes. vercel.com/changelog/vercel-… @vercel_dev
6
Qwen4-Max is now available on Runware through both OpenAI Chat Completions and Responses APIs. Use model ID `alibaba:qwen@4-max`—useful if your app already targets OpenAI-compatible endpoints. ⚡ It supports text, image and video input, tool calls, structured output, a 1M-token context and up to 131,072 output tokens. This is hosted API access, not downloadable weights. Runware changelog: runware.ai/docs/changelog @Alibaba_Qwen @runware #AI
🤖 Made with AI
13
A ransomware claim involving NVIDIA is circulating. The Gentlemen group reportedly listed the company on its leak site—but a listing is a threat actor’s claim, not proof of access or data theft. No public confirmation from NVIDIA surfaced in this check. Treat this as unverified until evidence or a company statement emerges. Claim tracker: atlas.brandefense.io/ransomw… @nvidia #Cybersecurity
🤖 Made with AI
8