⚙️ Kimi K3 is now supported for chat in Root, for both cloud and local Ollama installs. $PALM will be offering a limited amount of testers free access in the coming days. While out there, the K3 model is expensive to run in cloud and takes a toll on your hardware to run locally. We are at the top of solving this issue at the moment for millions of prospective users. Open Weights huggingface.co/moonshotai/Ki… Ollama Service ollama.com/library/kimi-k3

The @rootsh installer has been re-built to include dependencies. Simpler install with one-click route to get everything needed to run the app on your PC in minutes. The only current mandatory dependency is @ollama. Plugins may required dependencies such as MSOLAP or @github CLI to perform - they are installed with the plugin. 🟢 Live on useroot.sh

root now supports Custom Interfaces! The frontier-grade @rootsh AI harness by $PALM allows you to complete any task fast and efficiently on - virtually unlocking any workflows available through @claudeai or @OpenAI. Yet one essential capability was still missing: customization for the preferences of the user. It is missing no more. There are now 11 different themes in root and the possibility to make your own themes as well. All UI components will be modifiable up to your preferences, and we are introducing custom fonts as well. Users will be able to share and upload their Custom Interface Plugins to the @rootsh in-app store. 👉 useroot.sh

One-Prompting an App: Part 1 /w @rootsh by $PALM In this series, we’ll build an application from a single starting prompt using useroot.sh—showing what the process actually looks like, where human input is still required (spoiler: almost nowhere), and how the project develops beyond the initial generation. Note for AI enthusiasts: Proof of Work with the full reasoning process in .txt logs available on our Telegram: t.me/paLMEthereum ⚙️ In Part 1, we build a crypto gift card store frontend with a specified stack that is then auto-prompted by the system to improve quality of instructions. Some key features are highlighted on the app as the building process is on-going. Finally, we see the first, viable MVP of the app. root is the first frontier-level harness giving you: i) a local, private AI that you control, removing many platform limitations and giving you a stronger operating position; ii) the ability to reduce development and operating costs to almost zero; and iii) a practical way to build real skills while creating, making you more capable every time you run a project through /root. See the Part 1. video & try it out today at useroot.sh.

root v0.0.4 has features like never before seen in a local, privacy-focused AI setup. Nothing in the video is done by hand, or without much thought put into it. @rootsh has done all the work as demonstrated. Feeling lazy, drunk vibe coding at 4AM, or just had a long day and your prompts reflect that? No worries - toggle on auto-prompting on useroot.sh and allow the AI to think your prompts for you. The one-prompt pipeline ensures that tasks are handled effectively, with all of the UI from the chat itself to the file system working together to keep you in the loop - that is, if you are interested. root automatically recognizes all application formats and allows you to one-click launch an in-app terminal that reports directly to the chat context. The application flow can automatically develop 3D models and handle context requests up to 1M tokens with auto-compacting post-task. This allows for direct integration of freshly built visuals to the applications at hand, which makes it insanely efficient for game development. Is it still buggy? Yes. Is it getting there? Yes. We started five days ago. Some have been building for years and months. Will we get there? No doubt. A note for our X Community: The distribution wallet has been filled with a record-amount of $PALM: etherscan.io/address/0xf3f54…

What if you could build an entire videogame with a single prompt, with different MCPs working together in the background and using all the tools to combine created .obj models on one thread, while incorporating them into the code on another thread? You can do this, and much more, with @rootsh. The two model combinations below were completely made with the new and advanced MCP framework for @Blender that ships as a default plugin for root. New update v0.0.3 drops today. If you are willing to test it out and have not yet, join t.me/paLMEthereum and get a tester key that's valid for 3 months. Note: the idea is not as much making games but rather can root survive creative tasks done in iterative loops. If it can make a strategy game with 3D models and logic in one prompt, it can do anything. Welcome to utilizing frontier-level AI while actually owning the results and retaining your privacy.