/grok

Query xAI's Grok models directly from the command line.

Usage

/grok <query>
/grok --model <model> <query>
/grok --help

Arguments

Argument	Description
<query>	The question or prompt to send to Grok

Flags

Flag	Short	Description
--model	-m	Model to use (default: auto). Options: auto, fast, smartest, code, reasoning, cheap
--temperature	-t	Sampling temperature 0-2 (default: 0.7)
--help	-h	Show help message

Examples

Simple Query

/grok What are the best practices for error handling in Go?

Specify Model

/grok --model fast Explain what a mutex is

Code Analysis

/grok --model code Review this function for bugs: function add(a, b) { return a - b; }

Extended Reasoning

/grok --model reasoning Think through the trade-offs of microservices vs monolith

Lower Temperature for Factual

/grok --temperature 0.2 What is the time complexity of quicksort?

Model Aliases

Alias	Description	Best For
auto	Intelligent default (grok-4)	General queries
fast	Speed optimized (grok-4-fast)	Quick responses
smartest	Best quality (grok-4)	Complex analysis
code	Code optimized (grok-code-fast-1)	Programming tasks
reasoning	Extended thinking (grok-4.1-fast)	Multi-step reasoning
cheap	Most cost-effective (grok-4-fast)	Budget-conscious

Response Format

The response includes:

• Grok's answer
• Model used
• Token usage (input/output/total)
• Estimated cost in USD
• Response time

Related

• grok_query tool - Programmatic access with full parameters
• grok_models tool - List available models and pricing
• Using Grok skill - Natural language integration

Instructions

When the user runs /grok, execute the following:

1. Parse the command arguments:

   • Extract any --model or -m flag value (default: "auto")
   • Extract any --temperature or -t flag value (default: 0.7)
   • The remaining text is the query

2. If --help or -h is provided, show the usage information above

3. If no query is provided, ask the user what they'd like to ask Grok

4. Call the grok_query tool with:

   • query: The user's question
   • model: The specified model (or "auto")
   • temperature: The specified temperature (or 0.7)

5. Present Grok's response, including the metadata (model, tokens, cost)

6. If the query relates to code or technical topics the user is working on, offer to compare with Claude's perspective