# AddSearch MCP for AI Agents AI Agent Connect

> AddSearch lets you treat your website's search index like a live database. Your AI agent can query content with filters, push new documents, and pull real-time analytics on what users are actually searching for. It's the fastest way to audit site search performance and manage index content without jumping between dashboards.

## Overview
- **Category:** developer-tools
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_Lk90LaSuwxzHV4MxJBjoXa4PbfYg842s227Obcqs/ai-agent-connect
- **Tags:** site-search, nlp-search, search-analytics, api-indexing, content-discovery, search-ranking

## Description

Imagine you're trying to figure out why users aren't finding your products or why a specific category is underperforming. Instead of digging through complex analytics platforms and manually checking URLs, you can just ask your AI agent to do the heavy lifting. This Connector turns your AddSearch index into a conversational workspace. You can ask for the most common zero-result queries to see exactly what your customers want but can't find, or have the agent check if the right products are appearing for specific keywords. It handles the tedious parts of content management too, like pushing new JSON documents or clearing out old pages that are cluttering your results. Because it's available through the Vinkius catalog, you get this level of control over your search data directly within your favorite AI tools. It's about moving from searching for data to getting answers about your site's content and user behavior.

## Tools

### search_sorted
Run a search and organize the results by a custom variable. This helps you check the relevance of your sorting logic.

### stats_queries
Retrieve data on what users are typing into your search bar. This helps you identify content gaps and missing pages.

### search_keyword
Perform a standard keyword search across your indexed content. This mimics how a normal user would search your site.

### stats_clicks
Pull analytics on how many times users clicked on specific results. Use this to see which search results actually convert.

### autosuggest
Fetch autocomplete suggestions for specific search prefixes. This allows you to test your frontend search experience.

### delete_document
Permanently remove a document from your search index. This helps keep your results clean of outdated content.

### search_filtered
Find content using specific filters like category or brand. This lets you verify if specific products are ranking correctly.

### index_document
Add or update a document in your index using a JSON payload. Use this to push manual fixes or new content quickly.

### list_documents
Display a full list of all documents currently in your index. This is useful for auditing what the AI can actually see.

### search_pagination
Grab a specific page of results for testing navigation. Use this to see how your site handles multiple pages of results.

## Prompt Examples

**Prompt:** 
```
What are the most common things people are searching for that aren't showing any results?
```

**Response:** 
```
Based on the latest logs, here are the top queries with **zero results**:

| Query | Count |
| :--- | :--- |
| "internship 2026" | 142 |
| "sso integration" | 89 |
| "gdpr compliance pdf" | 55 |

These topics are high-interest but currently missing from your index. You might want to create content for these keywords.
```

**Prompt:** 
```
Check if our 'running shoes' are showing up correctly for the 'nike' brand.
```

**Response:** 
```
I ran a filtered search for **brand=nike** and **category=running shoes**. Here are the top 3 results:

1. Nike Air Zoom (Score: 0.98)
2. Nike Pegasus (Score: 0.94)
3. Nike Vaporfly (Score: 0.91)

The relevance ranking looks solid for these items.
```

**Prompt:** 
```
Give me the autocomplete suggestions for the word 'water'.
```

**Response:** 
```
The system returns these completions for the prefix "water":
- water resistant boots
- water bottle reviews
- water proof bags

The index is correctly populating suggestions for this entity.
```

## Capabilities

### Query content with custom filters
Filter your indexed content by categories, brands, or other custom variables in a single request.

### Push new documents via JSON
Add or update documents in your index by providing a JSON payload to your agent.

### Pull click-through analytics
Get real-time data on how many users actually clicked on your search results.

### Delete outdated index documents
Remove specific pages or documents from your index to keep your search results clean.

### Test auto-suggestion logic
Verify what autocomplete suggestions appear as users type into your search bar.

### List all indexed pages
See every document currently living in your search index for easy auditing.

## Use Cases

### Finding missing content
A content manager asks the agent to show the top 5 zero-result queries from last week to plan new blog posts.

### Fixing index errors
A developer has the agent use index_document to push a corrected JSON fix for a broken URL in the search results.

### Verifying category ranking
An e-commerce lead asks the agent to run search_filtered for brand=nike and category=shoes to check inventory visibility.

### Testing autocomplete logic
A UX designer asks the agent to test autosuggest for the prefix run to see if it correctly suggests running shoes.

## Benefits

- Identify content gaps by pulling zero-result queries via stats_queries to see what users want but can't find.
- Update your index instantly without manual uploads by using index_document to push JSON updates.
- Verify ranking accuracy for specific products by running search_filtered on categories like shoes.
- Test your frontend search experience using autosuggest to see what users see as they type.
- Clean up your index automatically by using delete_document to remove outdated or broken pages.
- Monitor user engagement by pulling click-through rates with stats_clicks to see which results actually convert.

## How It Works

The bottom line is you get a conversational interface for your entire site search index.

1. Connect your AddSearch account by providing your Site Key and Secret Key.
2. Link the Connector to your AI client through the Vinkius dashboard.
3. Ask your agent to perform searches, pull stats, or manage your index.

## Frequently Asked Questions

**Can I use AddSearch MCP to see what my users are actually typing?**
Yes. You can ask your agent to pull the latest search query analytics to see exactly what terms people are using most often.

**How does AddSearch MCP help with content planning?**
It helps you identify content gaps by pulling 'zero-result' queries, showing you exactly what your audience wants but can't find.

**Can the AddSearch MCP delete pages from my index?**
Yes. You can instruct your agent to permanently remove specific documents or outdated pages from your search index to keep results clean.

**Is AddSearch MCP good for checking search ranking?**
It's excellent for that. You can run filtered searches to verify that specific products or categories are appearing in the correct order.

**Can I push new content to my AddSearch index using an AI agent?**
Yes. You can have your agent push new content directly to your index using JSON data, skipping the need for manual dashboard uploads.

**Does AddSearch MCP provide click-through data?**
Yes. It can retrieve analytics on how many times users actually clicked on your search results, helping you measure content effectiveness.

**Can my AI agent manually index new pages?**
Yes. If you provide the AddSearch Secret Key, your agent can use the `index_document` tool. You can supply a URL and a JSON payload containing the title and body, and the agent will push it directly into your live search index without waiting for the web scraper.

**What kind of search analytics can I retrieve?**
Using the Secret Key, your agent can call two analytics endpoints: `stats_queries` (to see what users searched for, including top searches and zero-result queries) and `stats_clicks` (to see the click-through rates and popular URLs users navigated to from the search bar).

**Do I need the Secret Key if I just want to test search queries?**
No, if your goal is solely to run searches, paginate results, or check auto-suggestions, the public 'Site Key' is completely sufficient. The Secret Key is only required to list all documents, modify the index, and fetch analytics.