# Klevu (E-commerce AI Search) MCP

> Klevu AI Search MCP powers your e-commerce product discovery using natural conversation. Ask your agent to perform complex searches, audit category layouts, or fetch specific recommendations just by talking to it. You can execute keyword lookups, apply precise facet filters (like brand or size), and monitor global trending items without writing a single API call.

## Overview
- **Category:** ecommerce
- **Price:** Free
- **Tags:** site-search, product-discovery, merchandising, ai-search, recommendation-engine, search-relevancy

## Description

Connecting Klevu through this MCP lets you treat your e-commerce catalog like a conversation. Instead of building complex JSON payloads in Postman or digging through developer documentation, you talk to your agent about what you need—whether it's finding every jacket under $100, checking if a specific category page displays related items correctly, or seeing what products are currently spiking in popularity across the whole store.

It takes search logic and merchandising rules out of the code and into plain English. This level of control means you can quickly audit product rankings and recommendation setups for your site. By hosting this tool within the Vinkius catalog, you get immediate access to advanced discovery tools alongside any other e-commerce service your team uses.

## Tools

### search_autocomplete
Provides instant suggestions for users as they begin typing into a search bar.

### search_category
Retrieves product listings specifically configured for a given category page structure.

### search_filtered
Narrows down the catalog results by applying multiple specific criteria like color, size, and brand simultaneously.

### search_keyword
Searches the entire product catalog using a general keyword provided in plain text.

### search_pagination
Gets chunks of search results when you need to view more items on long result pages.

### search_product_id
Fetches the complete details for a single product using its unique catalog ID number.

### search_raw
Allows you to send custom, complex JSON payloads directly against the Klevu API endpoints.

### search_recs
Retrieves product suggestions based on machine learning models that predict what users might want.

### search_sorted
Performs a keyword search but allows you to specify how the results should be ordered (e.g., by price or date).

### search_trending
Shows the currently most popular and relevant products across your entire store.

## Prompt Examples

**Prompt:** 
```
Search for 'waterproof jackets' in my Klevu catalog
```

**Response:** 
```
Searching catalog… I've found 12 waterproof jackets. Top results include the 'Summit Pro Hardshell', 'RainGuard Trail Jacket', and 'Urban Explorer Parka'. Prices range from $89 to $245. Would you like to filter these by size or brand?
```

**Prompt:** 
```
Show me trending products for the 'Home Decor' category
```

**Response:** 
```
Retrieving trending items for 'Home Decor'… Current top sellers include 'Minimalist Ceramic Vase', 'Boho Woven Throw', and 'Smart Ambient Lamp'. These items are showing high engagement over the last 24 hours. Would you like to see visual similar items?
```

**Prompt:** 
```
Execute a filtered search for 'sneakers' with brand 'Nike'
```

**Response:** 
```
Filtering search for Nike sneakers… Found 8 matches. Featured models: 'Air Max Pulse', 'Dunk Low Retro', and 'Zoom Fly 5'. Would you like me to sort these by price from lowest to highest?
```

## Capabilities

### Conducting keyword searches
Find product listings across your entire catalog using natural language keywords.

### Filtering search results
Narrow down large result sets by applying specific characteristics like color, size, or brand.

### Checking category merchandising rules
Retrieve products configured for a specific category path to audit how your site displays content.

### Getting product recommendations
Fetch machine learning-driven suggestions, such as items frequently bought together or visually similar goods.

### Monitoring global trends
View the most relevant and fastest-selling products across your store to spot market opportunities.

### Running custom search payloads
Execute deeply nested, specific API queries using raw JSON structures.

## Use Cases

### Checking for product gaps during a seasonal launch
A merchandiser needs to confirm if their new fall collection is showing up correctly across multiple related categories. They ask, 'Show me all items in the Winter Outerwear category that are blue.' The agent uses `search_category` and `search_filtered` together to provide a precise list of what's currently visible.

### Debugging complex site search issues
A developer notices that 'waterproof jacket size 10' sometimes fails. Instead of writing multiple API calls, they ask the agent to run a targeted query using `search_filtered` and get a clean list, instantly confirming if the facet combination is supported.

### Analyzing competitor product positioning
A data analyst wants to see what products are currently gaining traction globally. They ask the agent to run `search_trending`, getting an immediate overview of top sellers and high-demand items that they can use for inventory planning.

### Building a recommendation engine prototype
A developer wants to test how product suggestions look. They ask the agent to execute `search_recs` on an existing search result, getting machine learning predictions without writing any backend code or managing external services.

## Benefits

- Bypass manual testing. Instead of repeatedly running separate queries for different filters, you can use the `search_filtered` tool to check complex combinations—like 'red shoes' *and* 'size 9'—all in one conversational prompt.
- Optimize site performance by monitoring what sells best right now. Use the `search_trending` tool to identify global product velocity and quickly pinpoint seasonal opportunities, avoiding guesswork about inventory needs.
- Control how products are displayed on your site. With `search_category`, you can audit whether a specific category page is correctly fetching all necessary related items according to your merchandising rules.
- Go beyond basic searches. If the AI needs deep data—say, comparing multiple product attributes simultaneously—the `search_raw` tool lets you execute complex JSON payloads without needing specialized coding knowledge.
- Improve user experience instantly. You can test how fast and accurate search results are by using `search_autocomplete`, ensuring that partial terms still guide users to the right products.

## How It Works

The bottom line is that this MCP turns complex product discovery into a simple conversation with your AI client.

1. Subscribe to this MCP and enter your Klevu Search URL and API Key.
2. Your AI agent uses the provided credentials to connect to your live e-commerce data.
3. You simply prompt your agent with a request, like 'Show me all blue shoes under size 10,' and it executes the search logic for you.

## Frequently Asked Questions

**How do I check if my product catalog supports complex filtering using Klevu AI Search MCP?**
You use the `search_filtered` tool. You just tell your agent what facets you want to combine, like 'color and size,' and it runs the query for you.

**Can I find out what products are selling well right now using Klevu AI Search MCP?**
Yes. Use `search_trending`. This tool shows current top sellers, letting you monitor global product velocity and spot seasonal spikes instantly.

**What if I need to run a query that the simple tools don't cover? Does Klevu AI Search MCP help?**
Absolutely. If your needs are highly specific, use `search_raw`. This tool lets you execute custom JSON search payloads against any deeply nested part of the Klevu API.

**How do I get product suggestions for a user on my site?**
You run `search_recs`. The agent uses this to fetch predictions based on machine learning, giving you suggested items like 'frequently bought together'.

**Does Klevu AI Search MCP handle product IDs for single lookups?**
Yes. If you know the ID number of a product, use `search_product_id`. This quickly retrieves all details for that single catalog item.