Azure.Search.Documents (.NET)

Build search applications with full-text, vector, semantic, and hybrid search capabilities.

Installation

dotnet add package Azure.Search.Documents
dotnet add package Azure.Identity

Current Versions: Stable v11.7.0, Preview v11.8.0-beta.1

Environment Variables

SEARCH_ENDPOINT=https://<search-service>.search.windows.net
SEARCH_INDEX_NAME=<index-name>
# For API key auth (not recommended for production)
SEARCH_API_KEY=<api-key>

Authentication

DefaultAzureCredential (preferred):

using Azure.Identity;
using Azure.Search.Documents;

var credential = new DefaultAzureCredential();
var client = new SearchClient(
    new Uri(Environment.GetEnvironmentVariable("SEARCH_ENDPOINT")),
    Environment.GetEnvironmentVariable("SEARCH_INDEX_NAME"),
    credential);

API Key:

using Azure;
using Azure.Search.Documents;

var credential = new AzureKeyCredential(
    Environment.GetEnvironmentVariable("SEARCH_API_KEY"));
var client = new SearchClient(
    new Uri(Environment.GetEnvironmentVariable("SEARCH_ENDPOINT")),
    Environment.GetEnvironmentVariable("SEARCH_INDEX_NAME"),
    credential);

Client Selection

Index Creation

Using FieldBuilder (Recommended)

using Azure.Search.Documents.Indexes;
using Azure.Search.Documents.Indexes.Models;

// Define model with attributes
public class Hotel
{
    [SimpleField(IsKey = true, IsFilterable = true)]
    public string HotelId { get; set; }

    [SearchableField(IsSortable = true)]
    public string HotelName { get; set; }

    [SearchableField(AnalyzerName = LexicalAnalyzerName.EnLucene)]
    public string Description { get; set; }

    [SimpleField(IsFilterable = true, IsSortable = true, IsFacetable = true)]
    public double? Rating { get; set; }

    [VectorSearchField(VectorSearchDimensions = 1536, VectorSearchProfileName = "vector-profile")]
    public ReadOnlyMemory<float>? DescriptionVector { get; set; }
}

// Create index
var indexClient = new SearchIndexClient(endpoint, credential);
var fieldBuilder = new FieldBuilder();
var fields = fieldBuilder.Build(typeof(Hotel));

var index = new SearchIndex("hotels")
{
    Fields = fields,
    VectorSearch = new VectorSearch
    {
        Profiles = { new VectorSearchProfile("vector-profile", "hnsw-algo") },
        Algorithms = { new HnswAlgorithmConfiguration("hnsw-algo") }
    }
};

await indexClient.CreateOrUpdateIndexAsync(index);

Manual Field Definition

var index = new SearchIndex("hotels")
{
    Fields =
    {
        new SimpleField("hotelId", SearchFieldDataType.String) { IsKey = true, IsFilterable = true },
        new SearchableField("hotelName") { IsSortable = true },
        new SearchableField("description") { AnalyzerName = LexicalAnalyzerName.EnLucene },
        new SimpleField("rating", SearchFieldDataType.Double) { IsFilterable = true, IsSortable = true },
        new SearchField("descriptionVector", SearchFieldDataType.Collection(SearchFieldDataType.Single))
        {
            VectorSearchDimensions = 1536,
            VectorSearchProfileName = "vector-profile"
        }
    }
};

Document Operations

var searchClient = new SearchClient(endpoint, indexName, credential);

// Upload (add new)
var hotels = new[] { new Hotel { HotelId = "1", HotelName = "Hotel A" } };
await searchClient.UploadDocumentsAsync(hotels);

// Merge (update existing)
await searchClient.MergeDocumentsAsync(hotels);

// Merge or Upload (upsert)
await searchClient.MergeOrUploadDocumentsAsync(hotels);

// Delete
await searchClient.DeleteDocumentsAsync("hotelId", new[] { "1", "2" });

// Batch operations
var batch = IndexDocumentsBatch.Create(
    IndexDocumentsAction.Upload(hotel1),
    IndexDocumentsAction.Merge(hotel2),
    IndexDocumentsAction.Delete(hotel3));
await searchClient.IndexDocumentsAsync(batch);

Search Patterns

Basic Search

var options = new SearchOptions
{
    Filter = "rating ge 4",
    OrderBy = { "rating desc" },
    Select = { "hotelId", "hotelName", "rating" },
    Size = 10,
    Skip = 0,
    IncludeTotalCount = true
};

SearchResults<Hotel> results = await searchClient.SearchAsync<Hotel>("luxury", options);

Console.WriteLine($"Total: {results.TotalCount}");
await foreach (SearchResult<Hotel> result in results.GetResultsAsync())
{
    Console.WriteLine($"{result.Document.HotelName} (Score: {result.Score})");
}

Faceted Search

var options = new SearchOptions
{
    Facets = { "rating,count:5", "category" }
};

var results = await searchClient.SearchAsync<Hotel>("*", options);

foreach (var facet in results.Value.Facets["rating"])
{
    Console.WriteLine($"Rating {facet.Value}: {facet.Count}");
}

Autocomplete and Suggestions

// Autocomplete
var autocompleteOptions = new AutocompleteOptions { Mode = AutocompleteMode.OneTermWithContext };
var autocomplete = await searchClient.AutocompleteAsync("lux", "suggester-name", autocompleteOptions);

// Suggestions
var suggestOptions = new SuggestOptions { UseFuzzyMatching = true };
var suggestions = await searchClient.SuggestAsync<Hotel>("lux", "suggester-name", suggestOptions);

Vector Search

See references/vector-search.md for detailed patterns.

using Azure.Search.Documents.Models;

// Pure vector search
var vectorQuery = new VectorizedQuery(embedding)
{
    KNearestNeighborsCount = 5,
    Fields = { "descriptionVector" }
};

var options = new SearchOptions
{
    VectorSearch = new VectorSearchOptions
    {
        Queries = { vectorQuery }
    }
};

var results = await searchClient.SearchAsync<Hotel>(null, options);

Semantic Search

See references/semantic-search.md for detailed patterns.

var options = new SearchOptions
{
    QueryType = SearchQueryType.Semantic,
    SemanticSearch = new SemanticSearchOptions
    {
        SemanticConfigurationName = "my-semantic-config",
        QueryCaption = new QueryCaption(QueryCaptionType.Extractive),
        QueryAnswer = new QueryAnswer(QueryAnswerType.Extractive)
    }
};

var results = await searchClient.SearchAsync<Hotel>("best hotel for families", options);

// Access semantic answers
foreach (var answer in results.Value.SemanticSearch.Answers)
{
    Console.WriteLine($"Answer: {answer.Text} (Score: {answer.Score})");
}

// Access captions
await foreach (var result in results.Value.GetResultsAsync())
{
    var caption = result.SemanticSearch?.Captions?.FirstOrDefault();
    Console.WriteLine($"Caption: {caption?.Text}");
}

Hybrid Search (Vector + Keyword + Semantic)

var vectorQuery = new VectorizedQuery(embedding)
{
    KNearestNeighborsCount = 5,
    Fields = { "descriptionVector" }
};

var options = new SearchOptions
{
    QueryType = SearchQueryType.Semantic,
    SemanticSearch = new SemanticSearchOptions
    {
        SemanticConfigurationName = "my-semantic-config"
    },
    VectorSearch = new VectorSearchOptions
    {
        Queries = { vectorQuery }
    }
};

// Combines keyword search, vector search, and semantic ranking
var results = await searchClient.SearchAsync<Hotel>("luxury beachfront", options);

Field Attributes Reference

Error Handling

using Azure;

try
{
    var results = await searchClient.SearchAsync<Hotel>("query");
}
catch (RequestFailedException ex) when (ex.Status == 404)
{
    Console.WriteLine("Index not found");
}
catch (RequestFailedException ex)
{
    Console.WriteLine($"Search error: {ex.Status} - {ex.ErrorCode}: {ex.Message}");
}

Best Practices

1. Use DefaultAzureCredential over API keys for production
2. Use FieldBuilder with model attributes for type-safe index definitions
3. Use CreateOrUpdateIndexAsync for idempotent index creation
4. Batch document operations for better throughput
5. Use Select to return only needed fields
6. Configure semantic search for natural language queries
7. Combine vector + keyword + semantic for best relevance

Reference Files

🧠 AGI Framework Integration

Adapted for @techwavedev/agi-agent-kit Original source: antigravity-awesome-skills

Hybrid Memory Integration (Qdrant + BM25)

Before executing complex tasks with this skill:

python3 execution/memory_manager.py auto --query "<task summary>"

Decision Tree:

• Cache hit? Use cached response directly — no need to re-process.
• Memory match? Inject context_chunks into your reasoning.
• No match? Proceed normally, then store results:

python3 execution/memory_manager.py store \
  --content "Description of what was decided/solved" \
  --type decision \
  --tags azure-search-documents-dotnet <relevant-tags>

Note: Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.

Agent Team Collaboration

• Strategy: This skill communicates via the shared memory system.
• Orchestration: Invoked by orchestrator via intelligent routing.
• Context Sharing: Always read previous agent outputs from memory before starting.

Local LLM Support

When available, use local Ollama models for embedding and lightweight inference:

• Embeddings: nomic-embed-text via Qdrant memory system
• Lightweight analysis: Local models reduce API costs for repetitive patterns