██████╗███╗ ███╗██████╗ ██████╗ ██╔════╝████╗ ████║██╔══██╗██╔══██╗ ██║ ██╔████╔██║██║ ██║██████╔╝ ██║ ██║╚██╔╝██║██║ ██║██╔══██╗ ╚██████╗██║ ╚═╝ ██║██████╔╝██║ ██║ ╚═════╝╚═╝ ╚═╝╚═════╝ ╚═╝ ╚═╝
Local-first, multi-agent AI coding in your terminal.
cmdr is an AI coding assistant that runs entirely on your machine using Ollama. No API keys, no cloud, no data leaves your laptop.
npm install -g cmdr-agent
cmdr
Published on npm with 1,800+ weekly downloads
More screenshots
Interactive model picker — choose from your locally available Ollama models on startup
Session dashboard with model info, permission mode, status bar, and operator boot sequence
Real-time context window tracking — see token usage, remaining capacity, and per-component breakdown
| Feature | ||
|---|---|---|
| 🔒 | Local-first | All inference on your hardware via Ollama |
| 🤖 | Multi-agent teams | Code review, full-stack, security audit presets |
| 🛠 | 13 built-in tools | Files, grep, glob, bash, git, web fetch, think |
| ✅ | Human-in-the-loop | Approve, deny, or always-allow each tool call |
| 🧠 | Context compaction | Multi-stage strategy keeps long conversations in bounds |
| 🔌 | Plugins & MCP | Extend with npm modules or Model Context Protocol servers |
| 💾 | Session persistence | Auto-save, resume, --continue |
| ↩️ | Undo | /undo reverts any file change the agent made |
| 📊 | Token tracking | /cost for per-session usage breakdown |
| 📁 | Project awareness | Reads CMDR.md for project-specific instructions |
# 1. Install Ollama — https://ollama.ai
ollama pull qwen3-coder:latest
# 2. Install cmdr
npm install -g cmdr-agent
# 3. Go
cmdr# One-shot mode
cmdr "fix the failing tests"
# Pick a model
cmdr -m llama3.1:8b
# Multi-agent review
cmdr --team review| Page | Description |
|---|---|
| Getting Started | Installation, first run, verify |
| Usage | CLI flags, slash commands, built-in tools |
| Multi-Agent Teams | Team presets and orchestration |
| Plugins & MCP | Plugin system and MCP integration |
| Configuration | Config files, env vars, permissions, CMDR.md |
| Benchmarks | Eval suite and HumanEval results |
| Architecture | Project structure and development setup |
See CONTRIBUTING.md.
MIT — Reyyan Ahmed