xmannii/nimruz-desktop — explained in plain English
Analysis updated 2026-05-18
Chat with an AI assistant in a Persian right-to-left desktop interface
Let an AI agent read, search, and edit files inside a linked project workspace
Use a ChatGPT account to access OpenAI's Codex model inside a desktop app
Save reusable experts, skills, and memories for recurring assistant tasks
| xmannii/nimruz-desktop | i4w7w4a/liquid_prnc_glass | qunabu/gravity | |
|---|---|---|---|
| Stars | 45 | 45 | 45 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 2/5 | 3/5 | 3/5 |
| Audience | vibe coder | developer | general |
Figures from each repo's GitHub metadata at analysis time.
macOS builds are unsigned, so Gatekeeper may block first launch until you allow it manually.
Nimruz Desktop is an open-source desktop app that gives you a chat assistant with a Persian, right-to-left interface, built for people who want to work with AI language models directly on their own files rather than only chatting in a browser tab. It is built with Electron, React, and the Vercel AI SDK, and works on Windows and macOS, with Linux support available if you build it yourself. Beyond plain chat, Nimruz can act as a workspace agent: you link a project folder, and the assistant can read, search, and write files, run shell commands, and produce artifacts and tasks inside that folder, asking for your approval before anything risky runs. You can connect it to a ChatGPT account to use OpenAI's Codex model, or bring your own key through OpenRouter or another compatible provider. When using Codex through a ChatGPT sign-in, the app runs it with a restricted setup that limits which files it can touch and blocks network access and several other tool types for safety. Other features include reusable saved experts and skills you can call up while chatting, a memory system that lets the assistant save and later forget specific facts, streaming responses with support for code, math, and diagrams, and light, dark, and system color themes. Everything is stored locally on your machine in an SQLite database, and provider API keys are kept in your operating system's secure keychain rather than in a plain text file. To try it from source you need Node.js version 22.12 or newer and the pnpm package manager, then you clone the repository, run pnpm install, and start it with pnpm dev. Pre-built installers for Windows and macOS are also published with each release, though the macOS build is not yet signed or notarized, so macOS may initially warn that the app looks damaged. The README does not state a software license for the project.
An open-source Persian-language desktop AI workspace agent that can chat, read and edit project files, and run tools with your approval.
Mainly TypeScript. The stack also includes TypeScript, Electron, React.
The README does not state a license for the project.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.