Quick Start
Installation
One-liner (recommended)
curl -sSL https://raw.githubusercontent.com/canoo/agent-nexus/main/install.sh | bash
Downloads the pre-built nexus binary and clones the repo to ~/.config/nexus.
After install, make sure ~/.local/bin is in your PATH:
export PATH="$HOME/.local/bin:$PATH"
Add this to your ~/.zshrc or ~/.bashrc to persist it.
From source
git clone https://github.com/canoo/agent-nexus.git
cd agent-nexus
bash setup-nexus.sh
Requires Go 1.25+ to build the TUI binary.
Launch the TUI
nexus
The interactive TUI walks you through the rest of setup:
⚡ NEXUS Framework Manager
v0.1.1
▸ Install NEXUS
Configure
Health Check
Uninstall NEXUS
j/k: navigate • enter: select • q: quit
| Screen | What it does |
|---|---|
| Install | Validates repo, creates symlinks, configures MCP, checks deps, pulls Ollama models |
| Configure | Edit Ollama host URL and model overrides, saves to .env |
| Health Check | Verifies Ollama reachability and symlink status |
| Uninstall | Removes all symlinks and the binary (with confirmation) |
What gets installed
Running Install NEXUS from the TUI creates symlinks so your AI CLIs read from NEXUS:
| Symlink | Points to |
|---|---|
~/.claude/CLAUDE.md | ~/.config/nexus/core/NEXUS.md |
~/.gemini/GEMINI.md | ~/.config/nexus/core/NEXUS.md |
~/.kiro/steering/nexus-orchestrator.md | ~/.config/nexus/core/NEXUS.md |
~/.claude/agents/ | ~/.config/nexus/personas/ |
Set up local LLM (optional)
Install Ollama then pull the default models:
ollama pull qwen2.5-coder:1.5b # supervisor band
ollama pull llama3.2:3b # logic band
Configure where Ollama runs:
cp .env.example .env
# Edit OLLAMA_HOST_URL if not using localhost:11434
Or use the TUI’s Configure screen.
Configure the MCP server
Add to your Claude Code config (~/.claude.json or the project mcp-configs/ templates):
{
"mcpServers": {
"nexus-ollama": {
"command": "node",
"args": ["~/.config/nexus/tools/mcp/server.mjs"]
}
}
}
Now Claude Code will automatically use ollama_commit_msg, ollama_lint_fix, and other tools for micro-tasks instead of cloud models.
What’s next
- Architecture — understand how routing works
- Personas — explore the built-in specialists
- Local Models — pick the right models for your hardware
- MCP Tools — configure and extend the delegation layer