Belmont banner
sinichi-edogawa sinichi-edogawa

Belmont

AI community

Description

Desktop app for running AI agents on your own files. Eight model providers, MCP servers, per-folder projects, and an approval gate in front of every dangerous action.

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

Belmont Tools

**Give an AI agent real tools. Keep the controls in your hands.**

Belmont Tools is a desktop app (Electron) for running AI agents directly on your own files, on your own machine. Eight model providers, MCP server support, per-folder projects, and a single approval gate in front of every dangerous action.

This app does not hide how the agent works — it gives you somewhere to watch it work, and a button to stop it.


Download (Windows)

File What it is
Belmont Tools Setup NSIS installer — installs permanently, appears in the Start Menu
Belmont Tools Portable A single .exe, no installation — carry it on a USB stick

Both links always resolve to the newest release, so they never go stale — and neither does a link you have shared elsewhere. The version number and the changelog live on the [Releases](https://github.com/sinichi-edogawa/belmont/releases) page.

**Requirements:**

  • Windows 10 or 11, 64-bit
  • An API key from one of the providers — or a Claude Code subscription, which needs no API key at all
  • Node.js, but only if you want to use MCP servers over stdio (they are launched with your system's npx). MCP servers over HTTP need nothing extra.

Settings and history live in `%APPDATA%/belmont-tools/`, so installing a new version over an old one erases nothing.


Why this app exists

The story is simple, and you may well have lived it yourself.

1. One model for everything is expensive — and unnecessary

The best coding agents today come with a subscription or an API bill that is not small. Yet not every job needs the most expensive model: tidying a README, renaming a variable, or explaining a single function can be handled just as well by something far cheaper — DeepSeek, GLM, Kimi, Gemini