Analysis Modes

Overview

Grok Search MCP Server provides two distinct analysis modes that determine the depth and structure of search results. The analysis_mode parameter controls whether you receive simple search results or comprehensive analytical insights.

Basic Mode
Comprehensive Mode

Basic Mode

Basic mode returns straightforward search results with minimal processing overhead. This mode is optimized for speed and simple queries where you need quick, factual information.

Characteristics

Token Limit: ~2,000 tokens maximum
Response Time: Fast (typically 2-5 seconds)
Caching: Not cached
Best For: Quick lookups, simple queries, real-time information

Response Structure

{
  "query": "your search query",
  "analysis_mode": "basic",
  "results": [
    {
      "title": "Article title or tweet content preview",
      "snippet": "Brief summary (max 280 chars for tweets)",
      "url": "Source URL",
      "source": "Source name or handle",
      "published_date": "YYYY-MM-DD",
      "author": "Author name"
    }
  ],
  "summary": "Brief overview of findings",
  "total_results": 10,
  "citations": ["url1", "url2"],
  "search_time": "2025-03-04T10:30:00Z"
}

System Prompt (Basic)

The server uses focused prompts for basic mode:

const basePrompt = `Provide current, accurate information about 
the following topic using live search data. Format your response 
as JSON with this exact structure:
{
"results": [
{
  "title": "Article title or tweet content preview",
  "snippet": "Brief summary",
  "url": "Source URL",
  "source": "Source name",
  "published_date": "YYYY-MM-DD",
  "author": "Author name"
}
],
"summary": "Brief overview of findings"
}`;

Example Usage

{
  "tool": "grok_search",
  "parameters": {
    "query": "latest tech news",
    "analysis_mode": "basic",
    "max_results": 5
  }
}

Comprehensive Mode

Comprehensive mode provides deep, structured analysis with timelines, quotes, multiple perspectives, and detailed context. This mode is designed for research, investigation, and understanding complex topics.

Characteristics

Token Limit: ~4,000 tokens maximum
Response Time: Slower (typically 10-20 seconds)
Caching: Automatically cached for 30 minutes
Best For: Research, analysis, complex topics, multiple perspectives

Response Structure

{
  "query": "original search query",
  "analysis_mode": "comprehensive",
  "comprehensive_analysis": "500+ word detailed analysis",
  "key_findings": [
    {
      "category": "main_story|development|context|impact",
      "title": "Specific finding title",
      "content": "Detailed explanation",
      "sources": ["url1", "url2"],
      "confidence": "high|medium|low"
    }
  ],
  "timeline": [
    {
      "date": "YYYY-MM-DD",
      "event": "Specific event description",
      "source": "Source name",
      "significance": "Why this matters"
    }
  ],
  "direct_quotes": [
    {
      "quote": "Exact quote",
      "speaker": "Speaker name and title",
      "context": "When and where",
      "source_url": "URL",
      "significance": "Why important"
    }
  ],
  "related_context": "Historical background",
  "multiple_perspectives": [
    {
      "viewpoint": "Perspective group",
      "content": "What this group thinks",
      "sources": ["url1"],
      "reasoning": "Why they hold this position"
    }
  ],
  "implications": {
    "short_term": "Immediate consequences",
    "long_term": "Long-term impacts",
    "stakeholders_affected": ["Group 1", "Group 2"]
  },
  "verification_status": {
    "confirmed_facts": ["Verified information"],
    "unconfirmed_claims": ["Claims needing verification"],
    "contradictory_information": ["Conflicting reports"]
  },
  "raw_results": [
    {
      "title": "Source title",
      "snippet": "Description",
      "url": "URL",
      "relevance_score": 8
    }
  ]
}

System Prompt (Comprehensive)

The comprehensive mode uses detailed analytical prompts:

const comprehensivePrompt = `You are a comprehensive research analyst. 
Provide deep, detailed analysis of the search topic using live search data.

CRITICAL INSTRUCTIONS:
Extract SPECIFIC details: exact dates, numbers, names, locations
Include DIRECT QUOTES with full attribution
Create a TIMELINE of events with precise dates
Analyze MULTIPLE PERSPECTIVES from different stakeholders
Provide HISTORICAL CONTEXT and background
Identify IMPLICATIONS and consequences
Verify information status and note contradictions
Be comprehensive but accurate - do not invent information`;

Example Usage

{
  "tool": "grok_news_search",
  "parameters": {
    "query": "climate policy changes 2025",
    "analysis_mode": "comprehensive",
    "max_results": 10,
    "from_date": "2025-01-01"
  }
}

Token Allocation

The implementation allocates different token budgets based on mode:

const requestData = {
  model: "grok-3-latest",
  messages: [...],
  max_tokens: analysisMode === "comprehensive" ? 4000 : 2000,
  temperature: 0.1,
  stream: false,
  search_parameters: searchParams
};

Mode Selection Guidelines

Use Basic Mode When

You need quick results
Query is simple and specific
Real-time data is critical
You want to minimize token usage

Use Comprehensive Mode When

Researching complex topics
Need multiple perspectives
Want timeline analysis
Require verified facts vs claims

Performance Comparison

Metric	Basic Mode	Comprehensive Mode
Max Tokens	2,000	4,000
Typical Response Time	2-5 seconds	10-20 seconds
Caching	No	Yes (30 min TTL)
Result Depth	Simple list	Multi-layered analysis
Citations	URLs only	Full metadata

Comprehensive analyses are automatically cached for 30 minutes to improve performance on repeat queries. Cache keys include all parameters to ensure accuracy.

Implementation Details

The mode affects response parsing logic:

if (analysisMode === "comprehensive" && parsedResults.analysis_mode === "comprehensive") {
  // Handle comprehensive analysis response
  finalResponse = {
    query: query,
    analysis_mode: "comprehensive",
    comprehensive_analysis: parsedResults.comprehensive_analysis || "",
    key_findings: parsedResults.key_findings || [],
    timeline: parsedResults.timeline || [],
    direct_quotes: parsedResults.direct_quotes || [],
    related_context: parsedResults.related_context || "",
    multiple_perspectives: parsedResults.multiple_perspectives || [],
    implications: parsedResults.implications || {},
    verification_status: parsedResults.verification_status || {},
    raw_results: parsedResults.raw_results || parsedResults.results || [],
    summary: parsedResults.summary || "",
    total_results: (parsedResults.raw_results || parsedResults.results || []).length,
    search_time: new Date().toISOString(),
    source: "grok-comprehensive-analysis",
    citations: citations,
    citation_metadata: citationMetadata
  };
}

All search tools support both analysis modes via the analysis_mode parameter. The default is "basic" if not specified.

Get Started

Core Concepts

Guides

Overview

Basic Mode

Characteristics

Response Structure

System Prompt (Basic)

Example Usage

Comprehensive Mode

Characteristics

Response Structure

System Prompt (Comprehensive)

Example Usage

Token Allocation

Mode Selection Guidelines

Use Basic Mode When

Use Comprehensive Mode When

Performance Comparison

Implementation Details

Build docs developers (and LLMs) love

Get Started

Core Concepts

Guides

​Overview

​Basic Mode

​Characteristics

​Response Structure

​System Prompt (Basic)

​Example Usage

​Comprehensive Mode

​Characteristics

​Response Structure

​System Prompt (Comprehensive)

​Example Usage

​Token Allocation

​Mode Selection Guidelines

Use Basic Mode When

Use Comprehensive Mode When

​Performance Comparison

​Implementation Details

Build docs developers (and LLMs) love

Overview

Basic Mode

Characteristics

Response Structure

System Prompt (Basic)

Example Usage

Comprehensive Mode

Characteristics

Response Structure

System Prompt (Comprehensive)

Example Usage

Token Allocation

Mode Selection Guidelines

Performance Comparison

Implementation Details