Command Flags Reference

All flags available in the /grok-x skill, organized by category.

X Search Flags

Control what X posts to search and how to filter them.

--handles

string

Comma-separated list of X handles to focus on. The @ symbol is optional.Maximum: 10 handlesConstraints:

Cannot be combined with --exclude
Each handle is automatically stripped of the @ prefix

Examples:

--handles elonmusk,xai
--handles @saylor,@jack  # @ is optional

--exclude

string

Comma-separated list of X handles to exclude from results. The @ symbol is optional.Maximum: 10 handlesConstraints:

Cannot be combined with --handles
This is an xAI API limitation — you must choose either inclusion or exclusion, not both

Examples:

--exclude openai,google
--exclude @spam,@bot  # @ is optional

--from

string

Start date for the search range. Must be in YYYY-MM-DD format.Format: YYYY-MM-DDExample:

--from 2026-02-01

--to

string

End date for the search range. Must be in YYYY-MM-DD format.Format: YYYY-MM-DDExample:

--to 2026-03-01

--images

boolean

Enable image analysis for posts containing images. Grok will analyze charts, screenshots, memes, and other visual content.Short form: -iDefault: falseImplementation: Adds enable_image_understanding: true to both x_search and web_search toolsExample:

--images
-i  # short form

--videos

boolean

Enable video analysis for posts containing videos. Grok will analyze video frames and content.Short form: -vDefault: falseConstraints:

Only works with x_search — this is an X Search exclusive capability
Has no effect when used with --web alone

Implementation: Adds enable_video_understanding: true to the x_search toolExample:

--videos
-v  # short form

Tool Flags

Activate additional capabilities beyond basic X search.

--web

boolean

Add web search alongside X search. Enables cross-platform research and fact-checking.Short form: -wDefault: falseImplementation: Adds web_search tool to the tools arrayUse cases:

Verify claims from X against broader web sources
Find corroborating evidence from news sites, blogs, and documentation
Get a complete picture across X and the open web

Example:

--web
-w  # short form

--analyze

boolean

Add code interpreter for quantitative analysis. Grok writes and executes Python code to compute statistics.Short form: -aDefault: falseConstraints:

Cannot be combined with --extract (xAI API limitation)

Implementation: Adds code_interpreter tool to the tools arrayUse cases:

Compute engagement statistics
Analyze posting frequency patterns
Calculate trend velocity
Generate quantitative reports from search results

Example:

--analyze
-a  # short form

--extract

string

Extract structured JSON output using a predefined schema. Automatically selects grok-4-1-fast-non-reasoning model.Short form: -xAllowed values: sentiment, narrative, timeline, profiles, claimsConstraints:

Cannot be combined with --analyze
Automatically overrides --fast and --deep flags
Forces model to grok-4-1-fast-non-reasoning (required for structured output with tools)

Implementation:

Sets text.format to json_schema with strict mode
Disables streaming
Returns guaranteed schema-compliant JSON

Example:

--extract sentiment
--extract timeline
-x claims  # short form

See Extraction Schemas for detailed schema definitions.

Model Flags

Control which Grok model to use for the query.

--fast

boolean

Use the fast non-reasoning model for quick lookups.Model: grok-4-1-fast-non-reasoningDefault: falseUse cases:

Quick lookups (“What did @naval post today?”)
Simple quote pulls
When latency matters more than depth

Trade-offs:

Significantly lower latency and cost
No extended reasoning chain
Good for straightforward queries

Example:

--fast

--deep

boolean

Use the flagship Grok model for maximum analytical depth.Model: grok-4-0709Default: falseUse cases:

Complex geopolitical analysis
Dense technical topics
Queries where the reasoning chain itself matters

Trade-offs:

Higher cost and latency
Maximum analytical capability
Not recommended for routine searches

Example:

--deep

If neither --fast nor --deep is specified, the model defaults to grok-4-1-fast-reasoning (or the value of GROK_X_DEFAULT_MODEL environment variable).When using --extract, the model is automatically set to grok-4-1-fast-non-reasoning regardless of these flags. This is a requirement of the xAI Responses API for structured output with tools.

Output Flags

Control how results are displayed.

--citations

boolean

Print all citation URLs after the response text.Short form: -cDefault: falseImplementation: Extracts URLs from response annotations and prints them in a formatted listExample output:

── Citations ──
  [SemiAnalysis] https://x.com/SemiAnalysis/status/189600234000000001
  [Reuters] https://www.reuters.com/technology/us-extends-chip-export-comment-period-2026-02-28/

Example:

--citations
-c  # short form

--no-stream

boolean

Disable streaming output. Response is buffered and printed all at once.Default: false (streaming is enabled by default)Use cases:

Debugging
When piping output to another command
When streaming causes display issues

Trade-off: You won’t see partial results as they arriveExample:

--no-stream

Flag Combinations

Valid Combinations

# X search + web search + image analysis + citations
/grok-x "query" --web --images --citations

# Handle filtering + date range + video analysis
/grok-x "query" --handles elonmusk,xai --from 2026-02-01 --to 2026-03-01 --videos

# Data analysis + web search
/grok-x "query" --analyze --web

# Structured extraction + handle filtering
/grok-x "query" --extract sentiment --handles saylor,jack

Invalid Combinations

The following combinations will fail with an error:

--handles + --exclude — xAI API limitation, choose one or the other
--analyze + --extract — mutually exclusive, choose quantitative analysis or structured output

Auto-Overrides

--extract automatically overrides model selection:When using --extract, the model is forced to grok-4-1-fast-non-reasoning even if you specify --fast or --deep. This is required by the xAI Responses API for structured output with tools.Example:

# Both of these use grok-4-1-fast-non-reasoning:
/grok-x "query" --extract sentiment
/grok-x "query" --extract sentiment --deep  # --deep is ignored

Environment Variables

These environment variables affect flag behavior:

XAI_API_KEY

string

required

Your xAI API key. Required for all queries.Format: Begins with xai-Configuration: Set in ~/.claude/settings.json

{
  "env": {
    "XAI_API_KEY": "xai-your-key-here"
  }
}

GROK_X_DEFAULT_MODEL

string

Default model when neither --fast nor --deep is specified.Default: grok-4-1-fast-reasoningValid values:

grok-4-1-fast-reasoning (recommended default)
grok-4-1-fast-non-reasoning
grok-4-0709

Configuration: Set in ~/.claude/settings.json

{
  "env": {
    "GROK_X_DEFAULT_MODEL": "grok-4-1-fast-reasoning"
  }
}

Get Started

Usage

Features

Reference

Command Flags Reference

X Search Flags

Tool Flags

Model Flags

Output Flags

Flag Combinations

Valid Combinations

Invalid Combinations

Auto-Overrides

Environment Variables

Build docs developers (and LLMs) love

Get Started

Usage

Features

Reference

​X Search Flags

​Tool Flags

​Model Flags

​Output Flags

​Flag Combinations

​Valid Combinations

​Invalid Combinations

​Auto-Overrides

​Environment Variables

Build docs developers (and LLMs) love

X Search Flags

Tool Flags

Model Flags

Output Flags

Flag Combinations

Valid Combinations

Invalid Combinations

Auto-Overrides

Environment Variables