# DeepOpinion MCP

> DeepOpinion MCP lets you run advanced Natural Language Processing (NLP) models through your AI agent without writing code. You can list all custom models available in your account, analyze single pieces of text for sentiment or topic, or process massive batches of data at once. It's immediate text intelligence accessible via natural conversation.

## Overview
- **Category:** developer-tools
- **Price:** Free
- **Tags:** nlp, text-analysis, no-code-ai, sentiment-analysis, machine-learning

## Description

You have messy text—customer reviews, support tickets, survey responses. Trying to get structured insights from that raw data usually means writing boilerplate code and calling APIs manually. This MCP changes the game. You connect it to any AI agent through Vinkius, giving your assistant immediate access to powerful NLP models trained in DeepOpinion.

Need to know the overall mood of 100 customer reviews? Use the batch processing tool for a quick count. Only analyzing one specific sentence before sending an email? Run a single prediction. If you're just starting out and need to see what kind of analysis is possible, use the model listing tool first.

It’s all about turning text into actionable data points—sentiment scores, topic classifications, or keyword extractions—all by simply asking your agent in plain language.

## Tools

### list_models
Retrieves a list of all custom DeepOpinion models you have trained and deployed.

### predict_batch
Processes multiple texts simultaneously, returning structured insights for every item in the list.

### predict
Runs an analysis on one specific text input using a chosen model ID.

## Prompt Examples

**Prompt:** 
```
List all my available DeepOpinion models.
```

**Response:** 
```
I've retrieved your models. You have 3 active models: 'Sentiment Analysis' (ID: sent-123), 'Topic Classifier' (ID: top-456), and 'Urgency Detector' (ID: urg-789). Which one would you like to use?
```

**Prompt:** 
```
Analyze this text using model sent-123: 'The new update is amazing and very fast!'
```

**Response:** 
```
Using the Sentiment Analysis model (sent-123), the result is: Positive (Confidence: 99.2%). The text expresses high satisfaction with performance and features.
```

**Prompt:** 
```
Run a batch prediction for model top-456 on these texts: ['How do I reset my password?', 'Where is my order?', 'I want a refund']
```

**Response:** 
```
Batch processing complete for model top-456:
1. 'How do I reset my password?' -> Category: Account Security
2. 'Where is my order?' -> Category: Shipping & Logistics
3. 'I want a refund' -> Category: Billing & Payments
```

## Capabilities

### See available models
Lists every custom NLP model you have access to within the DeepOpinion account.

### Analyze single text strings
Runs a prediction on one specific piece of text using a targeted model ID, giving an instant result.

### Process large text groups
Analyzes multiple texts at once (a batch), making it efficient for high-volume data review.

### Determine sentiment
Uses dedicated models to classify the emotional tone of any given text as positive, negative, or neutral.

### Categorize content topics
Assigns a specific topic or category (like 'Billing' or 'Shipping') to unstructured customer feedback.

## Use Cases

### Triage support tickets after a product launch
A Support Lead gathers 50 recent ticket summaries. Instead of manually reading them all, they prompt their agent to use predict_batch with the topic classifier model. The agent instantly returns a structured count: '30% Billing', '45% Login Issues', and '25% Feature Request'.

### Reviewing marketing copy for tone
A Product Manager inputs five different draft headlines into their agent. They ask to run a prediction using the sentiment analysis model, immediately seeing if the language sounds too aggressive or too weak.

### Validating a new ML pipeline step
A Data Scientist needs to test a newly trained 'Urgency Detector' model. They use list_models to confirm the ID and then run predict with one sample text to validate performance before scaling up.

### Analyzing competitor product descriptions
Someone copies three different competitor summaries. They prompt their agent to analyze them using a topic classifier model, allowing for direct comparison of which competitors focus on 'Sustainability' versus 'Price'.

## Benefits

- Stop writing complex API calls. You can trigger deep NLP analysis—like sentiment checks or topic classification—using simple, natural conversation commands.
- Handling massive data sets is easy. By using the predict_batch tool, you process hundreds of texts at once instead of sending them through one by one.
- You don't need to know model IDs upfront. Use list_models first to see exactly what custom models are available before running any prediction.
- Get immediate context on customer feedback. You can use the predict tool to analyze a single, critical comment and get instant insight into its tone or subject matter.
- The integration works across all major platforms. Connect your DeepOpinion account once through Vinkius and gain text intelligence in Claude, Cursor, Windsurf, and others.

## How It Works

The bottom line is, you talk to your AI agent like a person, and it handles all the complex text analysis underneath.

1. Subscribe to this MCP and enter your DeepOpinion API key in the Vinkius catalog.
2. Directly prompt your AI client, telling it which analysis you need (e.g., 'Analyze these texts for sentiment').
3. The agent calls the appropriate tool, runs the prediction, and returns structured insights directly into your chat.

## Frequently Asked Questions

**How can I see which NLP models are available in my account?**
You can use the `list_models` tool. Your AI agent will retrieve a complete list of all custom models you have trained or have access to in DeepOpinion.

**Can I process multiple sentences at once to save time?**
Yes! Use the `predict_batch` tool. It allows you to send an array of text strings to a specific `model_id`, making it perfect for analyzing large datasets quickly.

**What information do I need to run a prediction?**
To use the `predict` tool, you need the `model_id` (which you can find using `list_models`) and the `text` you want to analyze.

**How do I handle authentication when using the `list_models` tool?**
You must provide your DeepOpinion API Key during setup. This key authenticates your connection and ensures that any model listing or prediction request comes from your specific account.

**What happens if I use the `predict` tool with an invalid Model ID?**
The system will return a clear error message detailing the incorrect Model ID. You'll need to run `list_models` first to verify and correct the identifier before retrying the prediction.

**Are there any limits on how many texts I can use with `predict_batch`?**
The rate limit depends on your DeepOpinion subscription tier. For general usage, we recommend keeping batches under 100 items to maintain reliable performance and minimize potential throttling.

**Can the `predict` tool handle non-text inputs or only pure strings?**
The `predict` function is designed exclusively for string analysis. It requires a simple, clean text input that matches the expected data type of your selected NLP model.

**Does this MCP support all types of custom models I have trained in DeepOpinion?**
Yes, as long as the model is active and available within your DeepOpinion account, you can reference it via `list_models` and use its ID with the prediction tools.