AI & ML
Minke: lencx's Local-First Desktop Workspace for Your AI Agent
sun young Dev.to (EN Zone)
4 views
If the name lencx sounds familiar, it's probably because of his last project — the ChatGPT desktop client, the open-source app that put ChatGPT on thousands of desktops.
His new project is Minke (637 stars, Apache-2.0): a local-first desktop agent workspace powered by DeepSeek Harness.
The problem it solves
The pain of working with a coding agent is fragmented context. Your conversation lives in one window, your project files in a file manager, your terminal in another, and your browser tabs scattered everywhere. When the agent is mid-task and you want to see which file it touched, you juggle windows.
Minke pulls it into one workspace: work with an agent across conversations, project files, terminals, and visible browser tabs. Take over a page when you need to, review and edit the files the agent produced, or reach your workspace from another device.
In short: your agent stops being a black-box chat and becomes a collaborator that shares your context, your files, and your screen.
Three signals it's worth watching
Built by lencx — the ChatGPT-desktop author. He knows desktop apps, agent workflows, and what developers actually want; indie-polished tools are often more pleasant than big-company everything-apps.
Local-first — workspace, files, and data stay on your machine, no cloud dependency. For privacy-minded and enterprise users, that's a requirement, not a feature.
DeepSeek Harness underneath, but not locked to it — the project explicitly states it's "an independent community project, not an official DeepSeek product." You get an open, controllable workspace, not a shell tied to one model vendor.
The honest caveat
It's under active development — 637 stars, 69 forks, and the README warns that features, packaging, and local data schema may change. Don't expect a stable production tool today. It's also an Electron desktop app, and despite the "DeepSeek Harness" in the title, it's lencx's independent project — you run DeepSeek Harness locally or bring your own model.
I've localized the README and core docs to Chinese: https://github.com/yangshun2005/Minke-cn
If you find this project useful, a star on the original repo supports the author's ongoing maintenance.
Read original: https://dev.to/sun_young_517829fc09d0c05/minke-lencxs-local-first-desktop-workspace-for-your-ai-agent-4dbj
← Previous
How to Add a Verification Loop to Your AI Agent in 30 Minutes
Next →
Stop Overworking Your Code: A Guide to Debouncing
Related
Implementing AI Observability: Tracing LLM Calls End-to-End
AI & ML
1
DEV Community
Before you report LLM agreement, measure the human twice
AI & ML
1
DEV Community
How I Fine-Tuned a 7B LLM with LoRA and Unsloth
AI & ML
4
Dev.to (EN Zone)
How to Add a Verification Loop to Your AI Agent in 30 Minutes
AI & ML
2
Dev.to (EN Zone)
Comments0
No comments yet — be the first