Developers on an Apple Silicon Mac who want a single desktop app for coding agents; it helps to know what an API key and a shell command are, and to have Ollama if you want local models.

What can you build with radiant?

Ask an agent to fix a bug or explain a codebase in a project folder, approving each shell command as it goes. Run a local model through Ollama with no account or API key and keep your code on your machine. Build a loop that keeps retrying a step until your test command exits 0. Split a larger job into a graph so independent steps run in parallel under a spend budget. Switch between Claude, ChatGPT and a local model in one conversation to compare answers.

templetongroup /

radiant

A Mac app for chatting with coding agents on cloud or local models

114 stars
1 stars since 3 Oct
JavaScript
Updated yesterday
Some experience
Coding agents & dev tools

Open on GitHub (opens in a new tab) Website (opens in a new tab)

Radiant is a desktop app for working with coding agents on your own machine. You point a chat at a project folder, and the agent can read and write files and run shell commands there. By default it asks before each command, and every tool call shows up in an activity panel next to a real terminal.

It talks to many providers, including Anthropic, OpenAI, OpenRouter, xAI, Groq and Mistral, and to Ollama or LM Studio for local models. You can sign in with a subscription you already pay for, such as Claude, ChatGPT or GitHub Copilot, or paste an API key. Chats are stored in a neutral format, so you can switch models mid-conversation and keep your context.

Beyond single chats, it has a task board, loops and graphs. A loop is a list of steps where each step must pass a check, such as a command that exits 0, before the next begins. A graph runs steps in parallel unless one really reads another's output. It is built with Electron, React and an Express server, and the Mac app targets Apple Silicon.

What it does

Agent chat with tools and undo
Replies stream in with visible model thinking. You choose whether the agent asks before each command, auto-runs low-risk ones, or runs everything. Each reply that edits files leaves a checkpoint you can roll back without losing the conversation.
Local models checked against your Mac
Models you can download are labelled Runs well, Runs tight or Won't run, based on your machine's memory and cores, before you spend the download.
Tasks, loops and graphs
A kanban board moves cards through Queued, Working, Needs you, Review and Done. Loops make each step pass a check before moving on. Graphs run independent steps at the same time and report token and dollar cost.
Activity feed and real terminal
A side panel lists every tool call and its output, with a Preview tab for what the agent makes and a login shell built on node-pty and xterm.js.
Optional sandbox and local key storage
macOS can restrict commands to writing only inside the project folder, and optionally block the internet. API keys are kept in a local config file, and the server binds to 127.0.0.1 unless you turn on sharing.

What you could build

Ask an agent to fix a bug or explain a codebase in a project folder, approving each shell command as it goes.
Run a local model through Ollama with no account or API key and keep your code on your machine.
Build a loop that keeps retrying a step until your test command exits 0.
Split a larger job into a graph so independent steps run in parallel under a spend budget.
Switch between Claude, ChatGPT and a local model in one conversation to compare answers.

Get started

Download the signed Mac app from the releases page, or clone the repo and run it from source. Then start a session, point it at a project folder, and either let it use a running local model or add an API key in Settings.

npm install

npm run dev

Commands are copied from the project's README. Read it on GitHub for the full setup.