KotlinLLM is now open source! 🚀 It's a research prototype that lets you delegate logic to an LLM at runtime as explicit, persistent, and portable Kotlin code. The code evolves as your app encounters new scenarios during development and runs as plain Kotlin once generated. Learn more 👉 Explore it on GitHub 👉

We’re running our annual Kotlin Developer Survey and would love to hear from you. How’s Kotlin treating you lately? Help us improve the language, tools, and overall developer experience by sharing your feedback. It takes about 15 minutes: Looking forward to your insights! ☕

This product video was generated entirely by Claude Opus 5.5. Our colleague @kristel_tech gave it @HeyGen’s HyperFrames skills and pointed it at – out came this Kotlin promo. Try it in JetBrains Air or Junie.

🚀 Hiring: Senior Software Engineer — Poland (Remote) Build a new product from a blank repo in Rust at We’re looking for someone who loves: → Greenfield engineering → End-to-end ownership → Architecture & technical decisions → Mentoring 7+ years | Rust preferred | Go/Java/Kotlin considered | Claude Code required 📍Based in Poland? Apply:

📚 Auto-generate documentation for an entire repository, complete with diagrams. CodeWiki handles up to 1.4M lines across 8 languages and beats DeepWiki on quality (ACL 2026). Title: FSoft-AI4Code/CodeWiki URL: 📦 Overview CodeWiki is an AI-powered framework that automatically generates comprehensive documentation for whole software repositories. Beyond simple API references, it produces system-level docs with architecture diagrams, data-flow visuals, sequence diagrams, and prose. It tackles the challenge of keeping documentation current and complete across large, multi-language codebases. ❓ Challenges Solved Traditional documentation tools struggle with scale and context. CodeWiki addresses three core issues: documenting cross-module interdependencies, maintaining architectural context while scaling, and capturing system-level interactions rather than isolated components. 🛠 Methodology & Approach It runs in three stages. ・Hierarchical decomposition: algorithmic clustering inspired by dynamic programming partitions the codebase into coherent modules ・Recursive multi-agent processing: an adaptive agentic system delegates tasks dynamically for complex modules ・Multi-modal synthesis: it integrates text and visual artifacts (Mermaid diagrams), validating diagrams via Node.js ・It supports Python, Java, JavaScript, TypeScript, C, C++, C#, and Kotlin, and can run through Claude Code or Codex CLI with no per-token cost 🎯 Use Cases It fits generating living documentation for large multi-language codebases, onboarding material for new members, and interactive HTML output compatible with GitHub Pages. 📊 Experimental Results ・On its CodeWikiBench, quality scores were 79.14% for high-level languages (Python, JS) and 68.84% for managed ones (C#, Java) ・It improves on DeepWiki by +4.73 points and scored 82.45% on the 229K-line OpenHands project ・It has 1.2k GitHub stars, 193 forks, an ACL 2026 publication, and handles codebases from 86K to 1.4M lines ・The repository's own documentation was generated by CodeWiki, embodying its capability #AIAgents# #DevTools#

France’s Champagne output is set to fall by almost half from 2025 as scorching heat and drought batter yields. The national harvest would be one of the smallest in the past 30 years. Read more: 📷: Dave Kotinsky/Getty Images North America

🌾 The road into Dingo territory is now open. Crystal caravans are disappearing. Darklings are spreading. And what looked like a quiet road is starting to raise more questions than answers. Join Dino and Koalin as they head into the Thicket. Something is waiting deeper inside... 👀