# Lucidworks Fusion MCP

> Lucidworks Fusion provides full control over your corporate knowledge graph through natural conversation with your AI agent. Use this MCP to execute complex semantic searches, monitor machine learning ranking jobs, and update document indexes for deep enterprise discovery.

## Overview
- **Category:** knowledge-management
- **Price:** Free
- **Tags:** enterprise-search, semantic-search, machine-learning, vector-search, data-ingestion, search-ranking

## Description

This connector lets you take the complexity out of running an enterprise search platform. You can instruct your AI client to perform advanced queries that go beyond simple keyword matching—you'll run vector-based searches against specific documents or apps. Need to improve how relevant your search results are? Your agent handles sending user behavior signals, like clicks and conversions, directly into the system’s machine learning models for automatic ranking improvements. Furthermore, you can keep your data fresh by syncing brand new document mappings or auditing existing records in your physical search collections. Because Vinkius hosts this MCP, you connect once to access powerful tools designed specifically for Search Engineers and Data Scientists who need granular control over their infrastructure.

## Tools

### lw.list_collections
Lists the structured rules that are actively exporting billing data.

### lw.post_custom_query
Allows you to inject highly customized JSON logic that overrides standard Solr vector processing.

### lw.query_filtered
Performs a structural extraction of properties necessary for managing active accounts.

### lw.list_index_profiles
Identifies and lists the precise active arrays that handle native hold parsing.

### lw.index_documents
Performs an irreversible data validation process, extracting rich churn flags from documents.

### lw.list_jobs
Identifies and lists the precise active arrays related to native gateway authentication.

### lw.list_query_profiles
Runs an automated validation check that routes specific gateway history data.

### lw.query_search
Resolves precise AI vector rules matching strict profiles, identifying records within the platform's core system.

### lw.post_signal
Retrieves explicit cloud logging information while tracing vault limits.

### lw.query_sorted
Generates a highly available JSON payload with hard customer bindings, sorted by date descending.

## Prompt Examples

**Prompt:** 
```
Search the 'Support' app for 'password reset' using the 'default' profile
```

**Response:** 
```
Executing search in Fusion… I've retrieved 5 relevant documents. Top result: 'How to reset your corporate password' (Score: 0.95). Other matches include troubleshooting steps and security policies. Would you like to see the full document content?
```

**Prompt:** 
```
List all active ML training jobs for the 'Commerce' application
```

**Response:** 
```
I've identified 2 active jobs for 'Commerce': 1) 'Query-Intent-Model' (Status: Running, 45% complete), 2) 'Click-Re-Ranking' (Status: Waiting). I'll let you know once the intent model training is finished. Would you like to check the index profiles as well?
```

**Prompt:** 
```
Post a signal: user clicked on doc ID 'doc-987' in the 'Support' app
```

**Response:** 
```
Signal posted successfully. I've sent the click event for 'doc-987' to the Fusion signal pipeline for the Support app. This will be used to improve future search results for similar queries.
```

## Capabilities

### Run advanced semantic searches
Perform complex queries using both keywords and AI vectors against specific application profiles.

### Improve search relevance with user data
Send clickstreams or conversion signals to feed the system's machine learning models, making future searches better.

### Audit and manage document indexes
Update entire textual mappings or check which underlying search indices are active across your tenant.

### Monitor ML training jobs
Track the status of background data ingestion or machine learning model training to confirm everything is processing correctly.

### Inspect system configuration
List and audit how different query and index profiles are set up, allowing you to understand your search routing rules.

## Use Cases

### Debugging poor search ranking for a new feature
A Digital Experience Manager notices search results are missing key documents. They ask their agent to audit the query and index profiles, running `lw.list_query_profiles` first, then using `lw.query_filtered` to structurally extract properties, immediately pinpointing which data fields aren't being indexed correctly.

### Validating ML model performance after a traffic spike
A Data Scientist suspects the ranking model is outdated. They ask their agent to list active ML training jobs using `lw.list_jobs` and then send simulated user clicks via `lw.post_signal`. This confirms that the system is receiving fresh signals needed for accurate re-ranking.

### Onboarding a new data source into search
A Search Engineer needs to add a whole batch of new documents to the index. Instead of writing a bulk API call, they instruct their agent to use `lw.index_documents`, confirming that the process runs and extracts rich churn flags from the newly uploaded records.

### Troubleshooting data gaps in reporting
A team needs to see what collections are active for billing purposes. They ask their agent to execute `lw.list_collections`, which enumerates all attached structured rules, providing an immediate and clear view of the connected data sources.

## Benefits

- Stop logging into multiple dashboards to check search health. You can now list and audit underlying search indices and physical shards using a single command, giving you full visibility into your data distribution.
- You don't need to manually write complex API payloads for testing. Use the MCP to execute deep, custom JSON logic that overrides standard Solr vectors natively, making query debugging instantaneous.
- Improve relevance without manual model retraining. By using tools like `lw.post_signal`, your agent sends user behavior data (clicks, conversions) directly into Fusion's ML pipeline, improving search results automatically over time.
- Audit the rules governing your search logic from one place. You can list and inspect query profiles to understand exactly how AI models are configured in your routing layers, saving hours of manual documentation review.
- Track data integrity effortlessly. Monitor active ML training jobs or check index profiles directly through conversation, ensuring that critical background processes aren't failing silently.

## How It Works

The bottom line is you get to control high-level enterprise data architecture using only conversation, without writing a single API call.

1. Subscribe to this MCP on Vinkius.
2. Provide your Lucidworks Host URL and API Token credentials.
3. Use natural language commands through your AI client to execute search, index management, or job monitoring tasks.

## Frequently Asked Questions

**How do I check which document collections are active using Lucidworks Fusion MCP?**
You use the `lw.list_collections` tool with your agent. This command enumerates all explicitly attached structured rules, giving you a comprehensive list of every active data source for billing purposes.

**Can I check my ML job status using Lucidworks Fusion MCP?**
Yes, use `lw.list_jobs`. This tool identifies and lists the precise active arrays spanning native Gateway authentication, letting you confirm if your machine learning models are training correctly.

**What is the best way to improve search results with Lucidworks Fusion MCP?**
You should use `lw.post_signal`. This sends explicit cloud logging data, allowing you to feed user actions like clicks directly into the system for continuous improvement of search relevance.

**Does the Lucidworks Fusion MCP let me test custom queries?**
Absolutely. The `lw.post_custom_query` tool lets you inject deeply customized JSON logic that overrides Solr vectors natively, allowing for highly specific testing of your search parameters.

**I need to see all available index profiles in the Lucidworks Fusion MCP.**
Run `lw.list_index_profiles`. This tool identifies and lists the precise active arrays spanning native Hold parsing, giving you a map of your current indexing structure.