lencx/Minke

🐳 DeepSeek Harness Desktop

What it solves

Minke provides a local-first desktop workspace for AI agents, integrating conversations, file management, terminal access, and browser control into a single interface. It eliminates the need to switch between multiple tools when managing agent-driven tasks, while keeping sessions and settings stored locally on the user's computer.

How it works

Powered by the DeepSeek Harness (DSH), Minke operates as a desktop application (Electron-based) available for macOS, Windows, and Linux. It allows users to connect to either cloud-based LLMs or local models via providers like LM Studio and Ollama. The workspace features a split-view layout with a bottom panel for editing files and reviewing changes. It also includes a remote access feature that allows users to control their local workspace via a Web/PWA client on other devices, using secure tunnels like Tailscale or Cloudflare Access.

Who it’s for

It is designed for developers and power users who want a comprehensive, local-first environment to run AI agents that can interact with their local files, terminals, and the web browser.

Highlights

  • Shared Browser Control: Users can watch agents operate in visible browser tabs and intervene or provide annotated context.
  • Unified Workspace: Side-by-side integration of chat, files, and terminals with split views.
  • Multi-Device Access: Remote management of the local workspace via Web/PWA, and task initiation through WeChat, Telegram, and Discord.
  • Flexible Model Support: Compatibility with cloud providers, custom endpoints, and local model runners (Ollama, LM Studio).
  • Extensible Architecture: Support for DSH presets, skills, subagents, and native plugin management.

Related

  • Project
    ningbainb/deepseek-harness-desktop

    An open-source AI programming desktop client that integrates DeepSeek's DSH Runtime and Web UI into a full-featured workbench for Windows, macOS, and Linux.

  • Project
    hAcKlyc/MyAgents

    An open-source desktop workspace that integrates AI agents with local files, terminals, and a task management system to enable continuous, local-first AI workflows.

  • Project
    androoAGI/starnet

    StarNet is a local‑first desktop app that lets you build and run multiple AI agents in a pixel‑art space‑station UI. Each room represents a bounded team with its own memory, permissions and budget. Agents use real LLM providers (OpenAI, Anthropic, Google, OpenRouter, or a local Ollama model) and can call tools, send messages to chat platforms, run scheduled recipes, and output real files. The visual layout is the live workflow, and all state (transcripts, spend logs, memory) is persisted on disk with secrets kept in the OS keychain.

  • Project
    ayuayue/PiDeck

    PiDeck is an open‑source Electron‑based desktop workbench that wraps the pi coding‑assistant CLI (and DeepSeek Harness) to let developers manage multiple projects and AI coding‑agent sessions with integrated file, Git, terminal, and model‑configuration UI.

  • Project
    Aisland-SJL/dsh-worktable

    An agent-project workbench for DeepSeek Harness that provides a dockable multi-window workspace and a real-time control room for monitoring AI projects.