grok-insights - SKILL.md Agent Skill

name: grok-insights description: Search and analyze X/Twitter posts using the local Grok CLI for real-time social intelligence, sentiment analysis, and trend detection argument-hint: [query or topic] disable-model-invocation: true allowed-tools: - Bash - Read

Grok X Insights (Local CLI)

Get real-time social intelligence from X/Twitter using the locally installed Grok CLI. This skill uses the CLI built from this repository.

Arguments

$ARGUMENTS

A search query or topic to analyze (e.g., "artificial intelligence", "climate change")
If blank, will prompt for a topic

Prerequisites

The CLI is installed globally when you run npm install in this repo. Verify it's available:

grok --version

Ensure GROK_API_KEY is set in your environment or .env file.

Available Commands

1. Search Posts

Analyze X/Twitter posts about a topic with sentiment and themes.

grok search "QUERY" --time-window 4hr --limit 30 --analysis both -f text

Options:

--time-window or -t: 15min, 1hr, 4hr, 24hr, 7d (default: 4hr)
--limit or -l: Max posts to analyze, 1-50 (default: 50)
--analysis or -a: sentiment, themes, both (default: both)
-f or --format: Output format, json or text (default: json)

2. Analyze Topic

Deep analysis with custom aspects.

grok analyze "TOPIC" --aspects "sentiment,influencers,controversy" --time-window 24hr -f text

Options:

--aspects or -a: Comma-separated aspects to analyze
--time-window or -t: Time window for analysis
--limit or -l: Max posts to analyze

3. Get Trends

Identify trending topics and discussions.

grok trends --category technology --limit 30 -f text

Options:

--category or -c: technology, politics, sports, entertainment
--limit or -l: Max posts to analyze for trends

4. Chat with Grok

General chat, optionally grounded in X/Twitter data.

grok chat "What are people saying about AI?" --search -f text

Options:

--search or -s: Enable X/Twitter search
--search-limit or -l: Max posts to search if search enabled
--temperature: Response creativity 0.0-1.0 (default: 0.7)

Process

Step 1: Understand the request

Determine what kind of analysis the user wants:

Search: Specific topic analysis with sentiment/themes
Analyze: Deep dive with custom aspects
Trends: What's trending right now
Chat: General question, optionally with X data

Step 2: Run the appropriate command

Based on $ARGUMENTS:

For a simple search:

grok search "$ARGUMENTS" --time-window 4hr -f text

For trend detection:

grok trends -f text

For a grounded chat question:

grok chat "$ARGUMENTS" --search -f text

Step 3: Present findings

Summarize the results in a clear format:

Summary: Brief overview of what people are saying

Key Themes:

Theme 1
Theme 2

Sentiment: Overall positive/negative/neutral/mixed

Notable Points:

Observation 1
Observation 2

Citations: Include relevant X post URLs if available

Development

To test CLI changes locally:

# Build the CLI
npm run build

# Test commands
node dist/cli.js search "test query" -f text
node dist/cli.js trends -f text
node dist/cli.js chat "Hello" -f text

Example Usage

User: "What's the sentiment around Tesla stock?"

grok search "Tesla stock $TSLA" --analysis sentiment --time-window 24hr -f text

User: "What's trending in tech?"

grok trends --category technology -f text

User: "Analyze crypto with focus on sentiment and influencers"

grok analyze "cryptocurrency" --aspects "sentiment,key influencers,controversy" -f text