Edit

Azure MCP Server tools for Azure AI Search

The Azure Model Context Protocol (MCP) Server lets you manage Azure AI Search resources, including search services, indexes, knowledge sources, and knowledge bases with natural language prompts.

Azure AI Search (formerly Azure Cognitive Search) is a cloud search service that provides APIs and tools for building applications and agents that follow the Retrieval Augmented Generation (RAG) pattern. It supports vector and keyword retrieval, reranking, and agentic retrieval; for more information, see Azure AI Search documentation.

Note

Each Azure AI Search knowledge base exposes a native MCP endpoint for direct retrieval. For more information, see Call the MCP endpoint.

Note

Tool parameters: The Azure MCP Server tools define parameters for data they need to complete tasks. Some of these parameters are specific to each tool and are documented here. Other parameters are global and shared by all tools. For more information, see Tool parameters.

Get search index

This tool retrieves detailed information about Azure AI Search indexes, including index schema, fields, analyzers, scoring profiles, and other index properties. When you provide an index name, it returns properties for that index; without an index name, it returns all indexes in the specified service.

Example prompts include:

  • "Show me the details of index 'products-index' in Azure AI Search service 'my-search-service'."
  • "List all indexes in the Azure AI Search service 'enterprise-search'."
  • "What fields are in the 'users' index in service 'contoso-search'?"
Parameter Required or optional Description
Service Required The name of the Azure AI Search service (for example, my-search-service).
Index Optional The name of the search index within the Azure AI Search service. If you omit this parameter, the tool returns details for all indexes in the service.

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ❌ ✅ ❌ ❌

Query search index

This tool runs a search query against an Azure AI Search index and returns the matching documents and relevance metadata. Results typically include document fields, a relevance score, and any text highlights that match the query.

The tool automatically selects keyword-only or hybrid search based on the target index's schema. There's no parameter to force either behavior:

  • Keyword search: If the index has no vector fields, the tool sends your query text as a lexical (keyword) search and returns matches ranked by traditional text relevance.
  • Hybrid search: If the index has one or more vector fields that use a vector search profile with a configured vectorizer, the tool sends your query text as both a lexical query and a vector query. Azure AI Search runs both searches and merges the two ranked result sets using Reciprocal Rank Fusion (RRF) to produce a single combined ranking.

Hybrid search only works when the index defines vector fields, those fields reference a vector search profile, and the profile's vectorizer is fully configured (for example, an Azure OpenAI embedding deployment) so Azure AI Search can convert your query text to a vector at query time. If a vector field's vectorizer isn't configured or isn't reachable, the query can fail instead of silently falling back to keyword-only search.

This tool doesn't support requesting vector-only search: it always includes your query text as a lexical query alongside any vector queries, so you can't skip the keyword portion of a hybrid query. It also doesn't support multimodal (image or audio) vector search. The optional queryType and semanticConfiguration parameters let you choose the query mode: simple or full lexical query , or semantic ranking. They don't change whether the automatic hybrid vector search described here is applied.

Example prompts include:

  • "Search for 'machine learning' in the 'documents' index of my 'contoso-search' service."
  • "Query index 'products' for 'noise-canceling headphones' in Azure AI Search service 'fabrikam-search'."
  • "Search my 'content' index in 'my-search-service' for anything mentioning 'climate change'."
  • "Search the 'products-vectorized' index in 'contoso-search' for 'cozy waterproof hiking boots'."

If products-vectorized has vector fields with a configured vectorizer, this prompt triggers hybrid search: the tool returns results ranked by RRF, combining products whose text matches the keywords with products that are conceptually similar (for example, "insulated trail shoes") even without an exact keyword match.

Parameter Required or optional Description
Service Required The name of the Azure AI Search service (for example, my-search-service).
Index Required The name of the search index within the Azure AI Search service.
Query Required The search query to execute against the Azure AI Search index.
Query type Optional The query mode to use. simple uses the simple query mode, full (the default) uses the full Lucene query mode, and semantic applies semantic ranking in addition to any automatic hybrid vector search. semantic requires the index to have a semantic configuration.
Semantic configuration Optional The semantic configuration to use when Query type is semantic. If omitted, the index's default semantic configuration is used.

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ❌ ✅ ❌ ❌

Get knowledge base

This tool gets details for Azure AI Search knowledge bases. Knowledge bases encapsulate retrieval and reasoning capabilities over one or more knowledge sources or indexes. If you don't provide a knowledge base name, it returns details for all knowledge bases within the specified service.

Example prompts include:

  • "List all knowledge bases in the Azure AI Search service 'my-search-service'."
  • "Show me the knowledge bases in service 'contoso-search'."
  • "Get the details of knowledge base 'support-agent' in Azure AI Search service 'enterprise-search'."
Parameter Required or optional Description
Service Required The name of the Azure AI Search service (for example, my-search-service).
Knowledge base Optional The name of the knowledge base within the Azure AI Search service. If omitted, the tool returns details for all knowledge bases in the service.

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ❌ ✅ ❌ ❌

Retrieve from knowledge base

This tool executes a retrieval operation against an Azure AI Search knowledge base to find relevant information from its data sources. Provide either a single-turn query for retrieval or one or more conversational messages in role:content format. Specifying both query and messages isn't allowed.

Example prompts include:

  • "Run a retrieval with knowledge base 'support-agent' in Azure AI Search service 'my-search-service' for the query 'password reset steps'."
  • "Ask knowledge base 'product-docs' in search service 'contoso-search' to retrieve information about 'API rate limits'."
  • "Query knowledge base 'hr-policies' in search service 'hr-search' about 'vacation accrual policy'."
Parameter Required or optional Description
Service Required The name of the Azure AI Search service (for example, my-search-service).
Knowledge base Required The name of the knowledge base within the Azure AI Search service.
Query Optional Natural language query for retrieval when a conversational message history isn't provided.
Messages Optional Conversation history messages passed to the knowledge base. Each entry formatted as role:content, where role is user or assistant (for example, user:What policies apply to archived invoices?).

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ✅ ✅ ❌ ❌

Get knowledge source

This tool gets details of Azure AI Search knowledge sources. A knowledge source can point at an existing Azure AI Search index or represent external data (for example, a blob storage container) that Azure AI Search has indexed. Knowledge sources are used by knowledge bases during retrieval. If you don't provide a knowledge source name, it returns details for all knowledge sources in the specified service.

Example prompts include:

  • "List all knowledge sources in the Azure AI Search service 'my-search-service'."
  • "Show me the knowledge sources in the Azure AI Search service 'contoso-search'."
  • "Get the details of knowledge source 'product-index' in search service 'enterprise-search'."
Parameter Required or optional Description
Service Required The name of the Azure AI Search service (for example, my-search-service).
Knowledge source Optional The name of the knowledge source within the Azure AI Search service.

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ❌ ✅ ❌ ❌

List search services

This tool lists Azure AI Search services in a subscription and returns details for each service, including name, location, SKU, provisioning state, and endpoint.

Example prompts include:

  • "List Azure AI Search services in my subscription."
  • "What AI Search services do I have?"
  • "Show me my Azure AI Search resources."

Tool annotation hints:

Destructive Idempotent Open World Read Only Secret Local Required
❌ ✅ ❌ ✅ ❌ ❌

Related content

  • Last updated on