AIX

Run Claude Code with local AI models. No API keys. No cloud. Complete privacy.

Getting Started · Documentation · Contributing · Changelog

What is AIX?

AIX CLI is a bridge between local model servers and AI coding assistants. It connects LM Studio or Ollama to Claude Code — letting you use locally-running language models as the backend for your favorite AI dev tools.

No API keys. No cloud calls. No data leaving your machine.

┌──────────────────────────────────────────────────┐
│  $ aix-cli run                                   │
│                                                  │
│  ? Select model backend: Ollama                  │
│  ✔ Connected to Ollama                           │
│  ✔ Model selected: qwen2.5-coder:14b             │
│  ✔ Launching Claude Code...                      │
│                                                  │
│  Your code stays local. Always.                  │
└──────────────────────────────────────────────────┘

Why AIX?

• 🔒 Privacy-first — All inference runs locally on your hardware. Your code never leaves your machine.
• 🔑 No API keys — No subscriptions, no usage limits, no cloud dependencies.
• 🚀 GPU-accelerated — Take advantage of your local GPU for fast inference.
• 🔀 Single provider — Claude Code is the only supported AI coding assistant.
• ⚡ Zero config — Just run aix-cli run and start coding.

Compatibility notes

• LM Studio: Works natively with Claude Code via Anthropic-compatible API at /v1.
• Ollama: Use the --native flag to leverage Ollama's built-in ollama launch claude integration, which handles all API configuration automatically.

# Recommended for Ollama
aix-cli run --ollama --native -m qwen2.5-coder:14b

Getting Started

Prerequisites

| Requirement | Description | | -------------------------------------------------------------------- | ------------------------------------------ | | Node.js ≥ 18 | JavaScript runtime | | LM Studio or Ollama | Local model server (at least one required) | | Claude Code | AI coding assistant |

Install

npm install -g @iamharshil/aix-cli-cli

# Yarn
yarn global add @iamharshil/aix-cli-cli

# pnpm
pnpm add -g @iamharshil/aix-cli-cli

git clone https://github.com/iamharshil/aix-cli-cli.git
cd aix-cli
npm install
npm run build
npm link

Verify

aix-cli doctor

This checks that LM Studio / Ollama, Claude Code, and your environment are properly configured.

Usage

aix-cli run — Start a coding session

The primary command. Launches Claude Code backed by a local model.

# Interactive — prompts for backend and model
aix-cli run

# Specify backend and model
aix-cli run -b ollama -m qwen2.5-coder:14b
aix-cli run -b lmstudio -m llama-3-8b

# Use Ollama's native Claude Code integration (recommended for Ollama)
aix-cli run -b ollama --native -m qwen2.5-coder:14b

# Global shortcuts
aix-cli run --ollama --native -m gemma4

# Pass a prompt directly
aix-cli run -b ollama -m qwen2.5-coder:14b -- "Refactor auth middleware"

aix-cli init — Set up backend and model

Configure your preferred backend and load/select a model.

aix-cli init                                    # Interactive setup
aix-cli init -b ollama -m qwen2.5-coder:14b     # Ollama with specific model
aix-cli init -b lmstudio -m llama-3-8b -p claude # Full config in one command

aix-cli status — Check what's running

Shows status for both LM Studio and Ollama, including available and running models.

aix-cli status

aix-cli doctor — Infrastructure check

Checks that provider is running, port is accessible, model is available, and Claude config is correct.

aix-cli doctor

aix-cli setup — One-command setup

Quick setup for first-time users. Detects installed providers and configures defaults.

aix-cli setup                    # Interactive setup
aix-cli setup --provider ollama   # Use specific provider
aix-cli setup --force           # Overwrite existing config

aix-cli providers — Manage providers

List available providers or set a default.

aix-cli providers list              # Show providers with status
aix-cli providers set ollama        # Set default provider

aix-cli models — List models

Fetch and display available models from a provider.

aix-cli models list --provider ollama
aix-cli models list --provider lmstudio

aix-cli switch — Switch provider

Instantly switch providers without breaking Claude setup.

aix-cli switch ollama
aix-cli switch lmstudio

aix-cli disconnect — Disconnect

Remove connection cleanly.

aix-cli disconnect claude

aix-cli fix — Fix issues

Fix infrastructure issues - suggest starting backends, correct ports, reset config, fix model.

aix-cli fix

Command Reference

| Command | Aliases | Description | | ------------ | --------------- | ------------------------------------------------ | | run | r | Run Claude Code with a local model | | init | i, load | Set up backend, select model, configure provider | | status | s, stats | Show active provider, tool, endpoint, model | | doctor | d, check | Check infrastructure status | | setup | | One-command default setup | | providers | | List or set default provider | | models | | List available models | | switch | | Switch to a different provider | | disconnect | | Disconnect from provider | | fix | | Fix infrastructure issues | | update | upgrade, u | Update AIX CLI to the latest version | | config | c, settings | View, set, or reset CLI configurations |

Global Options

| Flag | Description | | ---------------------- | ------------------------------------------- | | -v, --version | Show version number | | -h, --help | Display help | | -b, --backend <name> | Model backend: lmstudio or ollama | | -m, --model <name> | Model name or ID to use | | -n, --native | Use Ollama's native Claude Code integration | | -V, --verbose | Show verbose output |

Configuration

AIX stores its configuration in the OS-appropriate config directory:

| Platform | Path | | -------- | ---------------------------------------- | | macOS | ~/Library/Application Support/aix-cli/ | | Linux | ~/.config/aix-cli/ | | Windows | %APPDATA%\aix-cli\ |

Config File

{
  "lmStudioUrl": "http://localhost",
  "lmStudioPort": 1234,
  "ollamaUrl": "http://localhost",
  "ollamaPort": 11434,
  "defaultTimeout": 30000,
  "defaultBackend": "ollama",
  "defaultProvider": "claude",
  "model": "qwen2.5-coder:14b"
}

Environment Variables

| Variable | Description | Default | | ---------------- | ---------------------------------- | ------- | | LM_STUDIO_PORT | Override the LM Studio server port | 1234 |

How It Works

        ┌───────────────────┐    ┌───────────────────┐
        │    LM Studio      │    │      Ollama        │
        │  (port 1234)      │    │  (port 11434)      │
        └────────┬──────────┘    └────────┬───────────┘
                 │                        │
            REST API                 REST API
                 │                        │
                 └───────────┬────────────┘
                             │
                    ┌────────┴──────────┐
                    │     AIX CLI       │
                    │  backend routing  │
                    │  model selection  │
                    │  config mgmt      │
                    └────────┬──────────┘
                             │
                             ▼
                    ┌──────────────┐
                    │  Claude Code │
                    │  --model X   │
                    └──────────────┘

1. LM Studio or Ollama runs a local inference server with an OpenAI-compatible API.
2. AIX CLI discovers available models, manages configuration, and orchestrates the connection.
3. Claude Code receives the model endpoint and runs as it normally would — except fully local.

Troubleshooting

1. Open LM Studio
2. Navigate to the Server tab (left sidebar)
3. Click Start Server
4. Confirm with aix-cli status

1. Install Ollama from ollama.com
2. Start the server: ollama serve
3. Pull a model: ollama pull qwen2.5-coder:14b
4. Confirm with aix-cli status

LM Studio: Open LM Studio → Search tab → download a model.

Ollama: Run ollama pull <model> to download a model (e.g., ollama pull llama3.2).

Then run aix-cli init to select and configure.

Check that the correct port is being used:

• LM Studio defaults to port 1234
• Ollama defaults to port 11434

You can configure custom ports in your AIX config file (path shown by aix-cli doctor).

Install Claude Code globally:

npm install -g @anthropic-ai/claude-code

Then re-run aix-cli doctor to confirm.

Security & Privacy

AIX is designed around a simple principle: your code never leaves your machine.

• ✅ All AI inference runs locally via LM Studio or Ollama
• ✅ No telemetry, analytics, or tracking of any kind
• ✅ No outbound network calls (except to localhost)
• ✅ No API keys or accounts required
• ✅ Fully open-source — audit the code yourself

Found a vulnerability? Please report it responsibly via our Security Policy.

Contributing

Contributions are welcome! See CONTRIBUTING.md for guidelines on how to get started.

git clone https://github.com/iamharshil/aix-cli-cli.git
cd aix-cli
npm install
npm run dev    # Run in development mode
npm test       # Run tests
npm run lint   # Lint

Related Projects

• LM Studio — Run local AI models with a visual interface
• Ollama — Run large language models locally
• Claude Code — Anthropic's AI coding assistant

License

MIT © Harshil