Getting Started
Nexum is an autonomous, developer-focused AI coding assistant designed to run local-first with Ollama, with failover to cloud reasoning models.
Prerequisites
- Node.js:
v22.0.0or higher - Ollama: Local instance running on
http://localhost:11434(orOLLAMA_API_KEYfor cloud) - Docker: For sandboxed execution of shell commands
Installation
1. Global Installation (Recommended)
Install globally from npm:
bash
npm install -g @nemesis-oss/nexum2. From Source
bash
git clone https://github.com/nemesis-oss/nexum.git
cd nexum
npm install
npm run build
npm link --forceBuilding the Docker Sandbox
Nexum isolates command execution inside a secure container. Build the sandbox image:
bash
docker build -t nexum-sandbox:latest docker/nexum-sandbox/Running Nexum
Navigate to any project directory and launch Nexum:
bash
nexumQuick Commands inside TUI
Ctrl+M: Open Model SwitcherCtrl+P: Open Universal Command Palette/doctor: Verify Ollama, models, language servers, and sandbox health/plan <task>: Plan and autonomously execute a multi-step objective/help: Open interactive cheat sheet