AX-Grok - Grok AI Coding Assistant CLI

An enterprise-grade AI coding assistant powered by xAI's Grok 4, providing terminal-based development with extended reasoning, vision capabilities, live web search, and Model Context Protocol (MCP) integration.

What This Skill Does

This skill guides you through using AX-Grok, a command-line AI coding assistant that leverages xAI's most capable models. It supports interactive coding sessions, headless automation, image analysis, real-time web search, and integration with 20+ specialized AI agents through the AutomatosX framework.

Installation & Setup

Follow these steps to install and configure AX-Grok:

1. **Install globally via npm:**

```bash

npm install -g @defai.digital/ax-grok

```

2. **Run initial setup:**

```bash

ax-grok setup

```

This will prompt you to enter your xAI API key, which will be stored encrypted in `~/.ax-grok/config.json`.

3. **Verify installation:**

```bash

ax-grok --version

ax-grok doctor

```

Available Models

AX-Grok supports multiple Grok models optimized for different use cases:

Usage Modes

Interactive Mode

Start an interactive coding session with full context and conversational interface:

```bash

Start new session

ax-grok

Resume previous conversation

ax-grok --continue

Start with specific model

ax-grok --model grok-4.1-fast

```

Headless Mode

Execute single commands without interactive prompts:

```bash

Analyze codebase

ax-grok -p "analyze this codebase structure and suggest improvements"

Fix errors in specific directory

ax-grok -p "fix all TypeScript errors" -d /path/to/project

Generate documentation

ax-grok -p "generate API documentation for all endpoints"

```

Essential Commands

Use these commands within interactive mode:

| Command | Description |

|---------|-------------|

| `/init` | Initialize project context (analyzes directory structure) |

| `/help` | Display all available commands |

| `/models` | Switch between available AI models |

| `/doctor` | Run diagnostics to check configuration and connectivity |

| `/search <query>` | Perform live web search |

| `/vision <path>` | Analyze image or screenshot |

| `/tools` | List all 17 built-in tools |

| `/mcp` | Manage Model Context Protocol integrations |

| `/agent <name>` | Launch specialized AutomatosX agent |

| `/todo` | View and manage task list |

| `/clear` | Clear conversation history |

| `/exit` | Exit the session |

Configuration

Configuration Files

Environment Variable

Alternatively, set API key via environment:

```bash

export XAI_API_KEY=your_xai_api_key_here

```

Project-Specific Setup

Create `.ax-grok/settings.json` in your project root:

```json

{

"model": "grok-4-0709",

"temperature": 0.7,

"reasoning_effort": "high",

"customInstructions": "Always write tests for new functions"

}

```

Key Features

Extended Reasoning

Enable deep reasoning mode for complex problems:

```bash

ax-grok -p "design a scalable microservices architecture" --reasoning-effort high

```

Vision Support

Analyze images, diagrams, or screenshots:

```bash

/vision screenshot.png

Then ask: "What UI improvements can you suggest?"

```

Live Web Search

Access real-time information during coding:

```bash

/search "latest xAI Grok API features 2026"

```

MCP Integration

Connect to 12+ production-ready MCP templates for extended capabilities. Templates include GitHub, PostgreSQL, Slack, filesystem access, and more.

AutomatosX Agents

Launch specialized agents for complex tasks:

Workflow Examples

Example 1: Project Analysis

```bash

ax-grok

> /init

> "Analyze this codebase and suggest performance improvements"

> /todo

> "Implement the top 3 recommendations"

```

Example 2: Bug Fixing with Vision

```bash

ax-grok

> /vision error-screenshot.png

> "This error appears in production. Find and fix the root cause"

```

Example 3: Research-Driven Development

```bash

ax-grok

> /search "best practices for React Server Components 2026"

> "Refactor this component using the latest best practices"

```

Part of AX CLI Ecosystem

AX-Grok is part of a family of AI coding assistants:

All share the same command structure and tool ecosystem.

Troubleshooting

Run Diagnostics

```bash

ax-grok doctor

```

This checks:

Common Issues

1. **API Key Not Found**: Run `ax-grok setup` to configure your xAI API key

2. **Connection Errors**: Check your internet connection and API key validity

3. **Model Access Denied**: Verify your xAI account has access to the requested model

4. **MCP Server Failures**: Use `/mcp status` to check server health

Best Practices

1. **Initialize projects**: Always run `/init` at the start of a session for better context

2. **Use custom instructions**: Create `.ax-grok/CUSTOM.md` for project-specific AI behavior

3. **Leverage reasoning mode**: Use high reasoning effort for architecture decisions

4. **Combine tools**: Mix web search, vision, and coding tools for comprehensive solutions

5. **Save context**: Use `--continue` to resume long-running projects

6. **Version control**: Add `.ax-grok/` to `.gitignore` except for `settings.json` if sharing config

Documentation & Resources

License

MIT License - Open source and free to use.

---

**Made by DEFAI Digital** - Enterprise AI solutions for developers.