# Azure AI Search MCP MCP

> Azure AI Search connects your agent directly to massive, private enterprise data stores in Azure. It lets you run both keyword searches and deep semantic vector queries against millions of documents without leaving your workflow. You can also audit index schemas and check synchronization health for full transparency into how your knowledge base is built.

## Overview
- **Category:** industry-titans
- **Price:** Free
- **Tags:** rag, information-retrieval, semantic-search, full-text-search, cloud-indexing, bm25

## Description

This MCP gives your AI client a direct line to complex, corporate knowledge bases hosted in Azure. Forget the limitations of simple database lookups; you can query everything from raw text documents to mathematical vector embeddings using one agent connection. You don't just get an answer; you inspect how that answer was generated by checking the source indexes and data pipelines themselves. The tool lets your agent perform precise lexical searches, but it also handles highly targeted relevance extraction across dimensional maps via vector search. It’s built for auditing: check which external systems are feeding data into your indexers and verify the structural schema of every component in place. When you connect this MCP through Vinkius, you get immediate access to all these specialized tooling capabilities from any compatible client.

## Tools

### get_index
Retrieves the specific configuration schema and details of a single Azure search index.

### list_datasources
Lists all external systems mapped to feed data into your Azure AI Search indexes.

### list_indexers
Provides a list of scheduled tasks that continuously synchronize data from source containers.

### list_indexes
Retrieves a complete listing of all available search indexes in the Azure environment.

### search_documents
Runs traditional full-text queries against your indexed documents using keywords.

### vector_search
Executes advanced similarity searches by comparing input vectors to highly dimensional embedded spaces.

## Prompt Examples

**Prompt:** 
```
Show me the configuration schema for our 'corporate-docs-v2' index.
```

**Response:** 
```
I've pulled the schema for 'corporate-docs-v2'. It contains 8 fields. The 'content_vector' field is mapped using HNSW cosine similarity. Semantic ranking is enabled. No language analyzers are explicitly overriding the default en.lucene settings.
```

**Prompt:** 
```
List the Azure Search indexers and tell me if any are failing.
```

**Response:** 
```
You have 2 indexers configured linking Blob containers. 'blob-sync-daily' is healthy on its schedule. However, 'cosmos-realtime-idxr' is reporting a configuration error — the destination container credentials appear disconnected on the backend.
```

**Prompt:** 
```
Run a full-text lexical search for 'Q3 Financial Audits' in the reports index.
```

**Response:** 
```
Searching the 'reports-index' via BM25 matching. I found 3 highly relevant paragraphs. The primary document `q3-audit.pdf` highlights a successful reconciliation process without anomalies. Would you like me to read the secondary findings?
```

## Capabilities

### Inspect all available indexes
Retrieve a list of every search index configured within your Azure environment.

### Check connected data pipelines
Get an explicit list detailing the external sources (like Blob Containers or SQL databases) supplying data to your indexers.

### Monitor sync jobs
List and audit all scheduled indexer tasks, checking their status and history.

### Run keyword searches
Execute traditional full-text queries against indexed documents using specific keywords.

### Perform semantic similarity searches
Identify contextually relevant passages by comparing input embeddings to stored vector data.

### View index structure details
Pull the exact schema, analyzers, and configuration settings for a single search index.

## Use Cases

### A Cloud Architect needs to prove compliance.
The architect must confirm that the customer records index isn't missing any data feeds. They use `list_datasources` to verify connections and then run `list_indexers` to check if all sync jobs are passing successfully.

### A Data Scientist needs specific context for a report.
Instead of getting too many general documents, the scientist uses `vector_search` with an embedding representing 'Q3 financial risk' and gets back only the three most semantically relevant paragraphs, citing their source index.

### A RAG Engineer is debugging a schema mismatch.
The engineer suspects the vector field was mapped incorrectly. They use `get_index` to pull the exact structural schema definition and confirm that HNSW cosine similarity is enabled before writing any new code.

### A Developer needs an index overview before deployment.
The developer wants a quick inventory of all available search capabilities. They call `list_indexes` to see every existing resource and then use `search_documents` with general keywords just to confirm basic connectivity.

## Benefits

- You don't just run a search; you audit the whole pipeline. Use `list_datasources` to check exactly where your unstructured data (like Blob Containers) is coming from, giving full provenance.
- Combine keyword and conceptual searches. Run standard text lookups with `search_documents`, then immediately follow up by using `vector_search` for deep semantic context retrieval on the same topic.
- Check index health instantly. Use `list_indexers` to see if your scheduled tasks are running on time, or use `get_index` to confirm the precise schema definition before a major query runs.
- Go beyond simple queries. The MCP allows you to pull the entire structure of your knowledge base by listing all available indexes with `list_indexes`, giving an immediate overview of capacity.
- Gain transparency into data flow. By calling `list_indexers` and seeing status updates, you instantly know if a critical corporate dataset is falling out of sync.

## How It Works

The bottom line is you can use your agent client to treat Azure AI Search as an extension of its own memory, giving it access to private corporate data.

1. Subscribe to this MCP on Vinkius and provide your Azure Search Endpoint and Admin Key.
2. Your agent client authenticates against the endpoint, granting read access to the knowledge base metadata.
3. You run a query—whether it's a keyword search or a vector similarity call—and get structured results detailing the source documents and schemas.

## Frequently Asked Questions

**How do I find out if my Azure AI Search indexes are up to date? (list_indexers)**
Run `list_indexers` first. This tool shows you which scheduled tasks run your data synchronization, telling you immediately if any indexer is reporting a configuration error or failing its schedule.

**What's the difference between using search_documents and vector_search? (vector_search)**
`search_documents` performs traditional keyword lookups, matching literal terms. `vector_search`, however, compares the meaning of your input against stored data embeddings for conceptual matches.

**I need to check the field types in my index. Which tool do I use? (get_index)**
Use `get_index` and specify the name of your search index. This pulls a detailed report on every field, including its analyzer type and if semantic ranking is enabled.

**How can I verify all possible data connections? (list_datasources)**
Call `list_datasources`. This tool reads the mapping configuration and lists every external system—like Azure SQL or CosmosDB containers—that feeds information into your search index.

**Can I list all existing indexes at once? (list_indexes)**
Yes, `list_indexes` retrieves a complete inventory of every single configured search index in the entire Azure AI Search environment. It’s useful for quick capacity checks.

**When I run `get_index`, what specific configuration details can I pull about a single search index?**
The tool returns the full structural schema for one index. You'll find explicit definitions for fields, including their data types, whether they support vector profiles, and analyzer settings like semantic ranking status.

**I need to verify exactly which cloud resources are mapped; how does `list_datasources` help?**
`list_datasources` pulls detailed REST maps showing the connection points. It tells you precisely where your indexers pull unstructured data from, listing all connected Blob Containers or Azure SQL databases.

**What is the best way to find out what indexes are available for querying before I run `search_documents`?**
`list_indexes` provides a complete list of every index configured in your environment. This lets you confirm the exact name and scope of the knowledge base you need to query.