# Doofinder MCP for AI Agents MCP

> Doofinder MCP gives your AI agents complete control over e-commerce search and catalog discovery. Use natural conversation to run complex keyword searches, apply deep filtering by properties like brand or color, predict product suggestions for partial queries, and audit performance analytics directly from the platform.

## Overview
- **Category:** ecommerce
- **Price:** Free
- **Tags:** site-search, search-analytics, predictive-search, conversion-optimization, e-commerce-tools

## Description

You don't have to jump between developer dashboards and spreadsheets just to test a new marketing campaign idea. This MCP connects your AI agent straight into the core of Doofinder’s e-commerce search engine. Instead of writing complex API calls for every scenario, you talk through it with natural language.

Your agent can perform deep keyword searches and then narrow those results down using structural filters—you tell it to show only 'red' items under $50 from a specific brand. Need to know if a certain product category is performing well? You can run automated checks on the search engine’s history or inspect raw catalog data, verifying exactly what your products are visible as. When you connect this MCP via Vinkius, your agent gains access to sophisticated tools that let you track click-through rates and test custom sort orders instantly. It's like having a dedicated e-commerce analyst sitting right next to your AI client.

## Tools

### search_custom
Runs specialized validation searches, helping you extract rich flags for specific business logic testing.

### search_filtered
Narrows down results by applying filters to properties like brand, color, or price range.

### get_search_engines
Sends an automated check of the gateway history to verify search engine status and connectivity.

### get_indices
Retrieves precise active arrays, giving you visibility into the system's overall index setup.

### get_items
Inspects deep internal data arrays to view specific product details and raw catalog information.

### get_stats
Captures key performance metrics, including click-through rates (CTR) and overall query volume history.

### search_keyword
Performs fundamental keyword searches across the entire headless platform.

### search_pagination
Retrieves detailed logging information, useful for tracing large volumes of search results and data limits.

### search_sorted
Generates structured product listings that sort by custom criteria like price or relevance.

### suggest
Provides predictive nodes to guide users with fast suggestions based on partial queries they type in.

## Prompt Examples

**Prompt:** 
```
I need to find all available running shoes, but only those that are blue and cost less than $100.
```

**Response:** 
```
**Search Results for Blue Running Shoes ($<100)**

*   **Brand:** SwiftStep
    *   Model: Nimbus Runner (Blue)
    *   Price: $89.99
*   **Brand:** Trailblazer
    *   Model: Alpine Sprint (Royal Blue)
    *   Price: $105.00 (Exceeded budget, excluded)

We found 4 matching items total. Would you like me to sort these by lowest price or best relevance?
```

**Prompt:** 
```
What are some popular search terms people are typing right now? Give me five suggestions.
```

**Response:** 
```
**Predictive Suggestions for Partial Query 'sun'**

Based on current trends, here are the top 5 suggested queries:

1.  `sunglasses`: (High volume)
2.  `sunscreen kit`: (Growing interest)
3.  `sunny day outfits`: (Seasonal)
4.  `sunrise yoga gear`: (Niche market)
5.  `sun-blocking curtains`: (Home goods)

Use these keywords to optimize your content strategy.
```

**Prompt:** 
```
Can you give me the search analytics for last week? I want to see total clicks and what my average CTR was.
```

**Response:** 
```
**Search Performance Report: Last 7 Days**

| Metric | Value |
| :--- | :--- |
| Total Queries Processed | 14,500 |
| Total Clicks Recorded | 1,250 |
| Average CTR | 8.6% |
| Query Velocity Trend | Up 7% from previous week |

The data shows a slight dip in weekend performance; focus marketing efforts on Tuesday mornings for better results.
```

## Capabilities

### Run advanced filtered product searches
Apply structural filters, specifying properties like color, brand, or price range to narrow down broad search results.

### Predict suggested keywords and items
Get fast predictive suggestions for partial queries, helping you quickly identify what customers might be looking for next.

### Audit product catalog indexes
Inspect deep internal arrays to sync un-cached raw catalog limits or check the structure of your entire product graph.

### Extract key performance metrics
Identify specific active data arrays spanning native hold parsing to capture exact click-through rates (CTR) and query velocity.

### Sort results by custom parameters
Generate JSON payloads that sort product lists according to hard customer bindings, like 'price:asc' or 'relevance:desc'.

## Use Cases

### A campaign needs testing across multiple categories
The marketer asks the agent: 'Run a search for 'summer footwear' and then narrow it down to only brands X and Y.' The agent uses `search_filtered` to deliver the precise, targeted result set instantly.

### Checking if an old product is still visible
The owner asks: 'Can you inspect the raw data for SKU 12345?' The agent uses `get_items` to pull deep internal arrays, confirming if the item exists and what its exact specifications are.

### Determining the best way to sort product pages
The team wants to compare default sorting vs. price-based sorting. The agent uses `search_sorted` with different parameters, providing a JSON payload for comparison in minutes.

### Diagnosing slow search performance spikes
A developer asks: 'Show me the query history and associated metrics for the last week.' The agent coordinates multiple tools, pulling data from `get_stats` and running a check using `get_search_engines`.

## Benefits

- Test complex search scenarios instantly. Instead of manual testing, you can use the `search_filtered` tool to test how results change when applying filters like brand or price range.
- Track performance metrics without leaving your chat window. Use `get_stats` to pull exact CTR and click-through data immediately for campaign validation.
- Deeply audit product visibility. By running `get_indices` and inspecting the catalog graph via `get_items`, you can verify if products are indexed correctly across all channels.
- Optimize user experience with predictive insights. The agent uses `suggest` to give you immediate ideas on popular partial queries, boosting site navigation.
- Control result presentation perfectly. Use `search_sorted` when you need results to always appear in a specific order, like price ascending or relevance descending.

## How It Works

The bottom line is, you stop translating business questions into technical API calls and start asking them directly to your AI client.

1. Subscribe to this MCP and provide your Doofinder Search Zone, HashID, and Management Token (API Key).
2. Your AI client authorizes the connection, giving your agent access to perform deep search queries and analytics checks.
3. Use natural conversation with your agent. Tell it what you need—'Show me all high-performing red items under $10.' Your agent executes the complex logic using specialized tools.

## Frequently Asked Questions

**How do I use Doofinder MCP to test if my product filters work correctly?**
You simply ask your agent to perform a filtered search. For example, 'Show me all green items under $50.' The tool handles the complex logic of combining multiple properties (color, price) into one clean result set, letting you prove your merchandising rules instantly.

**Can Doofinder MCP help me figure out what keywords customers are searching for?**
Yes. You can ask the agent to use predictive suggestion tools on partial queries like 'bath'. It returns a list of common, high-volume completions (e.g., 'bathroom rug,' 'bath towels'), giving you instant keyword ideas.

**What if I need to sort my search results by something other than relevance?**
You can use the dedicated sorting tool. Just tell your agent, 'Show me all hiking boots sorted by price, lowest first.' It provides a structured JSON payload that respects your custom ordering rules.

**Is Doofinder MCP better than just looking at my analytics dashboard?**
It's more dynamic. While dashboards show historical data, this MCP allows you to run live, targeted tests and get instant reports on things like current CTR or query velocity without logging into any external system.

**Do I need a developer to use Doofinder MCP for AI Agents?**
No. The whole point is that you don't. You talk to your agent using plain English, and the MCP translates those high-level questions into the specific API calls needed to get the data.