Sentiment Analysis MCP

Real MCP server for Claude Desktop. Analyze text sentiment with 6 emotion dimensions, classify intent with urgency, and batch process multiple texts. Local AI, zero API costs. Connect via Standby URL.

Features

• Sentiment Analysis: Detect text sentiment (positive, negative, neutral) with confidence score and 6 emotion dimensions (joy, anger, sadness, fear, surprise, disgust)
• Intent Classification: Classify text intent with urgency level (inquiry, complaint, request, feedback, purchase, cancel, support, etc.)
• Batch Processing: Analyze sentiment for multiple texts at once with individual results plus overall summary
• MCP Protocol: Fully compatible with Claude Desktop via HTTP+SSE
• Zero API Costs: Uses local AI server (Qwen 3.5)
• Pay-Per-Event: Charged only for actual usage

Prerequisites

• Node.js 18.0+
• Apify CLI: npm install -g apify-cli
• ntriq Local AI Server running at https://ai.ntriq.co.kr (or NTRIQ_AI_URL env var)

Installation

# Install dependencies
npm install

# Run locally
npm start

# Deploy to Apify
apify push

Tools

1. analyze_sentiment

Analyze text sentiment with confidence score and 6 emotion dimensions.

Input:

• text (string, required): Text to analyze
• language (string, optional): Text language code (default: en)
• max_tokens (number, optional): Max tokens for response (default: 100)

Output:

{
  "sentiment": "positive",
  "confidence": 0.92,
  "emotions": {
    "joy": 0.85,
    "anger": 0.05,
    "sadness": 0.02,
    "fear": 0.01,
    "surprise": 0.04,
    "disgust": 0.03
  }
}

2. classify_intent

Classify text intent with urgency level.

Input:

• text (string, required): Text to classify
• language (string, optional): Text language code (default: en)
• max_tokens (number, optional): Max tokens for response (default: 100)

Output:

{
  "primary_intent": "complaint",
  "urgency": "high",
  "sub_intents": ["service_issue", "billing_error"]
}

Intent Categories:

• inquiry (질문)
• complaint (불만)
• request (요청)
• feedback (피드백)
• purchase (구매)
• cancel (취소)
• support (지원)
• other (기타)

Urgency Levels:

• low
• medium
• high
• critical

3. batch_sentiment

Analyze sentiment for multiple texts at once.

Input:

• texts (array of strings, required): Array of texts to analyze
• language (string, optional): Language code for all texts (default: en)
• max_tokens (number, optional): Max tokens for response (default: 100)

Output:

{
  "results": [
    {
      "text": "I love this product!",
      "sentiment": "positive",
      "confidence": 0.95,
      "emotions": {
        "joy": 0.9,
        "anger": 0.01,
        "sadness": 0.01,
        "fear": 0.01,
        "surprise": 0.05,
        "disgust": 0.02
      }
    },
    {
      "text": "This is terrible",
      "sentiment": "negative",
      "confidence": 0.88,
      "emotions": {
        "joy": 0.02,
        "anger": 0.7,
        "sadness": 0.15,
        "fear": 0.05,
        "surprise": 0.02,
        "disgust": 0.06
      }
    }
  ],
  "summary": {
    "total_texts": 2,
    "positive": 1,
    "neutral": 0,
    "negative": 1,
    "average_confidence": 0.915,
    "dominant_emotions": {
      "joy": 0.46,
      "anger": 0.355,
      "sadness": 0.08,
      "fear": 0.03,
      "surprise": 0.035,
      "disgust": 0.04
    }
  }
}

Emotion Dimensions

Emotion	Description
Joy	Positive feelings, happiness, satisfaction
Anger	Frustration, irritation, displeasure
Sadness	Disappointment, unhappiness, dissatisfaction
Fear	Anxiety, worry, apprehension
Surprise	Unexpected reactions, novelty
Disgust	Aversion, distaste, repulsion

Claude Desktop Configuration

Add to your Claude Desktop config file (~/.claude/profiles/{profile}/config.json):

{
  "mcpServers": {
    "sentiment-analysis": {
      "command": "npx",
      "args": ["run", "start"],
      "cwd": "/path/to/sentiment-analysis-mcp",
      "env": {
        "NTRIQ_AI_URL": "https://ai.ntriq.co.kr",
        "PORT": "5000"
      }
    }
  }
}

Or use Standby URL for cloud-deployed Actor:

{
  "mcpServers": {
    "sentiment-analysis": {
      "url": "https://ntriqpro--sentiment-analysis-mcp.apify.actor/mcp?token={YOUR_APIFY_TOKEN}"
    }
  }
}

Pricing

Tool	Price	Notes
analyze_sentiment	$0.05	Per text analyzed
classify_intent	$0.05	Per text classified
batch_sentiment	$0.15	Per batch (multiple texts)

Pricing follows Apify's Pay-Per-Event model. You're only charged for successful API calls.

Example Usage

In Claude Desktop

I need to analyze the sentiment of these customer reviews:
1. "Absolutely love this product, exceeded all my expectations!"
2. "Terrible experience, worst purchase ever"
3. "It's okay, nothing special"

Use the batch_sentiment tool to analyze them and classify the intent of each.

Direct API Call

curl -X POST http://localhost:5000/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "analyze_sentiment",
      "arguments": {
        "text": "I absolutely love this!",
        "language": "en"
      }
    }
  }'

Legal Disclaimer

Analysis Notice: This service provides AI-generated sentiment and intent analysis. Results are statistical estimates and should not be used as the sole basis for business decisions, employment actions, or legal proceedings. Accuracy may vary by language, context, and text length. Text is processed in real-time and is not stored, retained, or used for model training. This service does not constitute professional psychological, linguistic, or business analysis.

Architecture

• Base Model: Qwen 3.5 (Apache 2.0 License)
• AI Server: ntriq Local AI (https://ai.ntriq.co.kr)
• Protocol: Model Context Protocol (MCP)
• Transport: HTTP + Server-Sent Events (SSE)
• Deployment: Apify Actor (Standby Mode)

Environment Variables

Variable	Default	Description
NTRIQ_AI_URL	https://ai.ntriq.co.kr	Base URL of sentiment analysis server
PORT	5000	Server port

License

Apache 2.0 (Qwen 3.5 Base Model)

Support

For issues or questions, contact ntriq engineering team or check the Apify documentation.

---

Last Updated: 2026-03-28