xAI Grok Integration - Autohand Docs

Overview

xAI's Grok models offer unique capabilities for code assistance. When integrated with Autohand, you get:

Access to Grok-2 and Grok-2 mini models
Real-time knowledge from X (Twitter) data
Large 128K context window
Vision capabilities for image understanding
Function calling support

API access: xAI API is available through console.x.ai. You'll need an xAI account with API access enabled.

Setup

Get started with xAI Grok.

Get your API key

Go to console.x.ai and sign in
Navigate to API Keys
Create a new API key and copy it

Configure Autohand

# Set environment variable
export XAI_API_KEY="xai-xxxxxxxxxxxxxxxxxxxx"

# Or configure via CLI
autohand config set xai.apiKey "xai-xxxxxxxxxxxxxxxxxxxx"

Verify your configuration:

# Start with xAI provider
autohand --provider xai --model grok-2

# Test with a prompt
autohand --prompt "Hello, which model are you?"

CLI configuration

Configure xAI in your ~/.autohand/config.json:

{
  "provider": "xai",
  "xai": {
    "apiKey": "${XAI_API_KEY}",
    "model": "grok-2",
    "maxTokens": 4096,
    "temperature": 0.7
  }
}

Configuration options

Option	Description	Default
`apiKey`	Your xAI API key	-
`model`	Model to use	`grok-2`
`maxTokens`	Maximum tokens in response	`4096`
`temperature`	Response randomness (0-2)	`0.7`
`baseUrl`	API endpoint	`https://api.x.ai/v1`

Available models

xAI offers Grok models optimized for different use cases.

Model	Context	Best for
`grok-2`	128K	Most capable, complex tasks
`grok-2-mini`	128K	Fast and cost-effective
`grok-2-vision`	128K	Image understanding

Switch models

# Set default model
autohand config set model "grok-2"

# Use during a session
/model xai/grok-2-mini

Best practices

Use environment variables: Keep API keys out of version control.
Choose the right model: Use grok-2 for complex tasks, grok-2-mini for quick responses.
Monitor usage: Check your xAI console for usage metrics.

Resources

Troubleshooting

Common issues

Issue	Solution
Invalid API key	Verify key at console.x.ai
Rate limited	Wait or check your rate limits
Model not available	Check model access in your account

Debug mode

# Enable verbose logging
AUTOHAND_DEBUG=true autohand --provider xai