Skip to main content

​
Overview

Intelligent enterprise search service powered by machine learning. Through Langdock’s integration, you can access and manage AWS Kendra directly from your conversations.
Authentication: API Key Category: Vector Database Availability: All workspace plans

​
Available Actions

aws_kendra.search
Searches your Kendra index using natural language queries Requires Confirmation: No Parameters:
  • query (TEXT, Required): Natural language query to search for in your Kendra index. For example: ‘How do I reset my password?’ or ‘sales report Q3 2024’
  • pageSize (NUMBER, Optional): Number of results to return per page. Maximum is 100. If not specified, defaults to 10
  • attributeFilter (MULTI_LINE_TEXT, Optional): JSON filter for results. Use ‘Get Index Configuration’ first to see available fields. OPERATORS: AndAllFilters, OrAllFilters, NotFilter. VALUE FORMAT depends on field type - Strings: direct value, Arrays: [“val1”, “val2”], Dates: {“DateValue”: “ISO8601”}, Numbers: {“LongValue”: 123}. Example:…
  • queryResultType (SELECT, Optional): Filter results by type. Choose what type of results to include
  • pageNumber (NUMBER, Optional): Page number to retrieve (1-10 for page size 10, 1-2 for page size 50). Note: AWS Kendra limits total retrievable results to 100
  • facets (MULTI_LINE_TEXT, Optional): JSON array of document attribute names to get facet counts. Simple format: [“_category”, “department”]. Advanced format with max results: [{“DocumentAttributeKey”: “_category”, “MaxResults”: 5}]
  • sortingConfiguration (MULTI_LINE_TEXT, Optional): JSON object to sort results. Format: {“DocumentAttributeKey”: “field_name”, “SortOrder”: “ASC” or “DESC”}. IMPORTANT: Use ‘Get Index Configuration’ first to verify which fields are sortable. Only fields marked as sortable in your index will work
  • spellCorrection (SELECT, Optional): Enable automatic spell correction for queries to improve search accuracy
  • userContext (MULTI_LINE_TEXT, Optional): JSON object for user-specific filtering. GenAI format: {“email_id”: “user@example.com”}. Standard format: {“Token”: “user-token”, “Groups”: [“group1”, “group2”], “UserId”: “user123”}
  • visitorId (TEXT, Optional): Unique identifier for tracking user sessions (e.g., a GUID). Do not use personally identifiable information like email
  • requestedDocumentAttributes (MULTI_LINE_TEXT, Optional): JSON array of document attribute names to include in response (max 100). Common fields: [“_document_title”, “_source_uri”, “_created_at”, “_excerpt”, “department”]. Reduces response size by limiting fields
  • collapseConfiguration (MULTI_LINE_TEXT, Optional): JSON object to group/collapse similar results. Basic: {“DocumentAttributeKey”: “_version”}. With expansion: {“DocumentAttributeKey”: “_version”, “Expand”: true, “ExpandConfiguration”: {“MaxResultItemsToExpand”: 10, “MaxExpandedResultsPerItem”: 3}, “MissingAttributeKeyStrategy”: “COLLAPSE”}
  • documentRelevanceOverrides (MULTI_LINE_TEXT, Optional): JSON array to boost specific fields/values. Format: [{“Name”: “field_name”, “Relevance”: {“Importance”: 1-10, “ValueImportanceMap”: {“value1”: 10, “value2”: 5}, “Freshness”: true, “Duration”: “1d”, “RankOrder”: “ASCENDING”}}]
Output: Returns the operation result

​
Get index configuration

aws_kendra.describeIndex
Returns field configuration for your Kendra index. Shows field names, types (STRING, DATE, LONG), and properties (searchable, sortable, facetable). Run this BEFORE searching to know which fields you can use for filtering and sorting. Requires Confirmation: No Parameters: None Output: Returns the operation result