# Constructor MCP MCP

> Constructor MCP lets your AI agent manage product discovery end-to-end. Instead of manually testing site search or navigating complex category dashboards, you ask for it conversationally. It handles everything from ML-ranked searching to applying strict filters (size, color) and generating personalized recommendations. You get a full audit trail of how products are found online—all through chat.

## Overview
- **Category:** ecommerce
- **Price:** Free
- **Tags:** site-search, product-discovery, personalization, recommendation-engine, nlp, e-commerce-optimization

## Description

You can take complete control of your site search experience using this MCP. Stop running manual tests on backend dashboards just to verify if product discovery works correctly. With this connection, you tell your agent what you need—like 'Show me all the waterproof jackets in size large' or 'What should we recommend for a first-time buyer?' The tool handles the complex queries and returns structured data instantly. You can audit everything: check how deep a category tree goes, see exactly which brands are prioritized, or get recommendations based on simulated user behavior. Since this MCP lives within the Vinkius catalog, you connect your agent once and gain access to all these critical e-commerce tools without switching platforms.

## Tools

### autocomplete
Predicts available search terms as you type a query into the agent.

### browse_brand
Retrieves products by inspecting deep arrays associated with a specific brand.

### browse_category
Lists all available products within an entire product department or category hierarchy.

### browse_collection
Identifies and retrieves pre-set groups of marketing items or static collections.

### custom_search
Executes a generalized search across the product catalog based on your input.

### search_filtered
Runs a search and restricts the results list to match specific attributes, like size or color.

### search_pagination
Checks how search results behave across multiple pages of output.

### get_recommendations
Pulls personalized suggestions for products using defined filtering models.

### search_products
Finds products by accessing the core record set within the platform.

### search_sorted
Lists product results based on structured rules, such as best sellers or price order.

## Prompt Examples

**Prompt:** 
```
Search for 'running shoes' in Constructor
```

**Response:** 
```
Searching for 'running shoes'... I found 120 products. The top ML-ranked items are 'Ultra-Light Runner' ($120) and 'Trail Master v2' ($145). Would you like to apply any filters like size or brand?
```

**Prompt:** 
```
What products are recommended in the 'home-page-trending' pod?
```

**Response:** 
```
Retrieving trending recommendations... I found 5 items including 'Eco-Friendly Yoga Mat' and 'Wireless Noise-Cancelling Headphones'. These are currently trending based on your global collaborative filtering model.
```

**Prompt:** 
```
Browse the 'Outdoor Furniture' category
```

**Response:** 
```
Browsing 'Outdoor Furniture' (ID: group_789)... I found 45 products in this category hierarchy. Top sub-categories include 'Patio Sets' and 'Garden Chairs'.
```

## Capabilities

### Search Products by Query
Run general product searches using ML ranking based on keywords and user intent.

### Filter Results by Attributes
Refine search results immediately, restricting the list to specific colors, sizes, or features.

### Browse Product Categories
Navigate deep product directory trees and see all available products within a defined department.

### Get Personalized Suggestions
Pull dynamic recommendations using collaborative filtering models based on simulated user activity.

### Check Brand Inventories
Inspect deep arrays of products belonging to a specific manufacturer or brand line.

### Audit Curated Collections
Retrieve predefined marketing clusters and static product groups for promotional review.

## Use Cases

### Debugging Size Availability
A QA tester needs to confirm that only red, size medium items show up. They run the agent and instruct it to 'Search for blue jackets' but add a filter using `search_filtered` for color: red and size: M. The system immediately verifies the correct subset of products.

### Auditing Brand Visibility
An E-commerce Manager wants to ensure that all items from 'Brand X' show up when a user is browsing outdoor gear. They use `browse_brand` for 'Brand X' within the 'Outdoor Gear' category, confirming every product ID is active and visible.

### Checking Pagination Logic
A Product Owner suspects that search results break down on page three. Instead of clicking through manually, they run `search_pagination` to force a structured check across multiple pages instantly, finding the exact breakdown point.

## Benefits

- Stop wasting time manually checking search results. Use `search_filtered` to instantly test how product lists behave when you apply strict rules like size or color.
- Never guess what a customer will look for again. The agent uses `autocomplete` to predict popular terms as you type, making your debugging faster.
- Audit marketing efforts easily. Instead of checking dashboard reports, use `browse_collection` to verify that every curated product cluster is correctly listed and rankable.
- Improve recommendation accuracy instantly. Run the `get_recommendations` tool to simulate user behavior and see exactly which products should be suggested on a given page.
- Verify site structure without limits. Use `browse_category` to map out entire departmental trees, ensuring your product taxonomy is deep and functional for every client type.

## How It Works

The bottom line is you can audit and test all of your site's product discovery logic without ever leaving the conversation window.

1. Subscribe to this MCP and enter your Constructor.io Public API Key into the Vinkius platform.
2. Your AI agent connects to the toolset, allowing it to execute complex e-commerce queries through natural language prompts.
3. You receive structured data containing product listings, filtered results, or recommended items directly in your chat interface.

## Frequently Asked Questions

**How do I test specific color and size combinations using search_filtered?**
You tell your agent you want to filter results by attributes. The tool uses `search_filtered` to restrict the product set, allowing you to check if only red, size 10 items appear, regardless of other criteria.

**Is Constructor MCP better than just using search_products?**
Yes. While `search_products` finds general products, this MCP gives you the specialized tools to audit *how* those products are found—for example, checking if they are part of a specific brand line using `browse_brand`.

**What is the difference between browse_category and search_products?**
A category browsing tool like `browse_category` maps out entire departments (e.g., 'Home Goods'). `search_products` performs a query against all records, regardless of their primary department.

**Can I use the get_recommendations tool for marketing?**
Absolutely. You can simulate user paths and run `get_recommendations` to see which products should be promoted or featured in a specific marketing pod, validating your merchandising strategy.

**How do I get started with `search_products` to test my e-commerce workflow?**
You first need to connect your Constructor.io API key in the MCP setup interface. This gives your agent the credentials it needs to run any search command against your live catalog data.

**When should I use `search_pagination` instead of just running a general product search?**
Use this tool when you anticipate needing results beyond the initial page load. It automatically handles validation checks, so your agent can reliably fetch and process deep catalog data without hitting rate limits.

**Can `browse_category` provide all available attributes for filtering?**
Yes, it generates a detailed JSON payload mapping the full taxonomy structure. This lets you map out every possible attribute and sub-classification within that category, which is crucial for advanced filtering logic.

**What happens if I use `autocomplete` with an ambiguous search term?**
The tool extracts properties driving active account logic. It helps your agent narrow down the user’s intent by suggesting precise categories or brand names, making subsequent searches much more accurate.

**Can my agent check the ML ranking for a specific product search?**
Yes. Use the 'search_products' tool. The agent will retrieve results ranked by Constructor's ML engine, allowing you to audit how products are surfaced based on specific keywords and intent signals.

**How do I retrieve personalized recommendations via the agent?**
Provide the 'pod_id' to your agent and use the 'get_recommendations' tool. The agent will query the collaborative filtering models to return a list of products tailored to your specified recommendation logic.

**Can I test attribute filtering like color or size through chat?**
Absolutely. The 'search_filtered' tool allows you to pass exact attribute mappings (e.g., 'color:blue,size:L'). Your agent will verify how the API restricts results to those specific structural bounds.