cmdr

Local-first AI coding assistant for VS Code.
Your models. Your machine. Your data.

cmdr brings local AI coding assistance directly into VS Code — no cloud, no API keys, no data leaving your machine. Powered by Ollama and the cmdr agent framework.

Features

@cmdr Chat Participant — Ask questions, generate code, and get help right inside VS Code's built-in chat panel
Inline Completions — Real-time AI code suggestions as you type, powered by Ollama's fill-in-the-middle (FIM) models
Smart Code Actions — Fix errors, explain code, refactor, and write tests from the lightbulb menu or command palette
Code Review — Get AI-powered review of your uncommitted git changes
Multi-Model Support — Switch between any Ollama model on the fly
Zero Config — Auto-starts cmdr serve in the background; just install and go

Requirements

Ollama installed and running locally
cmdr installed globally:
```
npm install -g cmdr-agent
```

Quick Start

Install Ollama and pull a model:
```
ollama pull qwen3-coder:latest
```
Install cmdr:
```
npm install -g cmdr-agent
```
Install this extension from the VS Code Marketplace
Open any project — the extension auto-starts cmdr serve in the background
Use @cmdr in the chat panel or run commands from the palette

Settings

Setting	Default	Description
`cmdr.model`	`qwen3-coder`	Model for chat and code actions
`cmdr.completionModel`	`qwen2.5-coder:7b`	Model for inline completions
`cmdr.effort`	`medium`	Effort level (low / medium / high / max)
`cmdr.ollamaUrl`	`http://localhost:11434`	Ollama server URL
`cmdr.inlineCompletions`	`true`	Enable inline completions
`cmdr.autoStart`	`true`	Auto-start cmdr serve
`cmdr.port`	`4200`	Server port

Commands

Command	Description
`cmdr: Open Chat`	Open the `@cmdr` chat panel
`cmdr: Explain Selection`	Explain the selected code
`cmdr: Refactor Selection`	Refactor the selected code
`cmdr: Write Tests`	Generate tests for selected code
`cmdr: Fix Error`	Fix diagnostics in the current file
`cmdr: Review Changes`	AI review of uncommitted git changes
`cmdr: Switch Model`	Change the active Ollama model

Why cmdr?

100% Local — Everything runs on your machine. No telemetry, no cloud calls.
Open Source — MIT licensed. Fork it, extend it, make it yours.
Fast — Lightweight extension that connects to a local cmdr server.
Model Agnostic — Use any model Ollama supports: Qwen, Llama, DeepSeek, Gemma, and more.

License

This extension is licensed under the MIT License.

cmdr - Local AI Coding Assistant

sinisterchill