Experimental MCP Service for Azure AI Search

This experimental MCP service provides you with the following capabilities:

- Retrieve a List of AI Search Indices from your Service
- Get Details about the Schema and Configuration of a Specific AI Search Service
- Create new Indices
- Update Existing Indices
- Run Queries against specific AI Search Indices
- Create and Update indexers

Important Disclaimers

We are still implementing and testing out these capabilities.

Some of the tool groups described above are not yet available but would be coming soon.

We are continuously working on adding and upgrading the service capabilities. More tools are going to be added in the near future.

These tools have the ability to modify data in your AI Search Service.

Please note that all tools are currently marked experimental and may change behavior without advanced notice.

You should always review and verify all AI-generated content for accuracy and correctness.

Please proceed with caution and use at your own risk.

Available Tools

For the time being, the following tools are available from the service:

| Tool Name | Tool Group | Tool Description |
|-----------------------------------------|---------------------|----------------------------------------------------------------------------------|
| list_index_names | READ_INDEX | Retrieve all names of indexes from the AI Search Service |
| list_index_schemas | READ_INDEX | Retrieve all index schemas from the AI Search Service |
| retrieve_index_schema | READ_INDEX | Retrieve the schema for a specific index from the AI Search Service |
| create_index | WRITE_INDEX | Creates a new index |
| modify_index | WRITE_INDEX | Modifies the index definition of an existing inde |
| delete_index | WRITE_INDEX | Removes an existing index |
| add_document | WRITE_DOCUMENTS | Adds a document to the index |
| delete_document | WRITE_DOCUMENTS | Removes a document from the index |
| query_index | READ_DOCUMENTS | Searches a specific index to retrieve matching documents |
| get_document_count | READ_DOCUMENTS | Returns the total number of documents in the index |
| list_indexers | READ_INDEXER | Retrieve all names of indexers from the AI Search Service |
| get_indexer | READ_INDEXER | Retrieve the full definition of a specific indexer from the AI Search Service |
| create_indexer | WRITE_INDEXER | Create a new indexer in the Search Service with the skill, index and data source |
| delete_indexer | WRITE_INDEXER | Delete an indexer from the AI Search Service by name |
| list_data_sources | READ_INDEXER | Retrieve all names of data sources from the AI Search Service |
| get_data_source | READ_INDEXER | Retrieve the full definition of a specific data source |
| list_skill_sets | READ_INDEXER | Retrieve all names of skill sets from the AI Search Service |
| get_skill_set | READ_INDEXER | Retrieve the full definition of a specific skill set |
| fk_fetch_local_file_contents | FETCH_FILE_CONTENTS | Retrieves the contents of a local file path (sample JSON, document etc) |
| fk_fetch_url_contents | FETCH_FILE_CONTENTS | Retrieves the contents of a URL (sample JSON, document etc) |

MCP Service Tool Groups

We do not want your MCP Host to be overwhelmed with the amount of tools coming from this service.

These are the available tool groups and their purposes:

- READ_OPERATIONS - tools used for read-only operations
- WRITE_OPERATIONS - tools that are used for creating, modifying or removing entries. Alias for ALL
- READ_INDEX - tools used to list and describe the indices in the service
- WRITE_INDEX - tools used to Create, delete, update, or configure indices
- READ_DOCUMENTS - tools used to query the indices to retrieve documents
- WRITE_DOCUMENTS - tools for Adding, Updating or Deleting documents from an index
- WRITE_INDEXERS - tools used to configure indexers, data sources & skill sets
- READ_INDEXERS - tools used to retrieve information about data sources, skill sets and indexers

Pre-Requisites

You will need to have configured a Service Principal that will be used to authentic against the AI Search service you are interacting with. You may also use the AI Search Key with read/write privilleges on the index dependening on what capabilities you are looking to leverage from the service.

Python 3.12 or later is needed, and you have to install uv as well to leverage the service.

The repository below contains a Terraform script that can accelerate the provisioning of an AI Search service and service principal necessary to help you get started.

Installing Dependencies

Follow the links below to install uv, python and the module containing the MCP service

- Installing uv
- Installing Python if Necessary with uv

You can install the mcp service as follows:

````bash

uv build

pip install dist/mcp_server_azure_ai_search_preview-0.3.1-py3-none-any.whl