Release Notes

Note: SDK v2.0 is compatible with Amorphic v3.4 and above

Features (19)

  1. Automated Reasoning Policies – Full lifecycle management for Amazon Bedrock Automated Reasoning policies has been introduced. Users can create, list, retrieve, update, and delete policies, and manage build workflows end to end: start a build to upload policy documents, list and inspect workflows, export build results, cancel an in-flight build, delete a completed workflow, and retrieve the resulting policy definition.

  2. AI Project Comments and Stewards – AI Projects now expose collaboration and governance methods in the generated client, including listing, adding, and deleting project comments and adding or removing project stewards under the same owner and editor guardrails used by the platform.

  3. Access Token Lifecycle – Programmatic management of personal access tokens has been added. Users can create, list, update, and delete tokens through dedicated Access Tokens APIs, enabling automation without embedding long-lived credentials in application code.

  4. Resource Access Requests – A request-and-approve workflow for resource access is now available. Callers can create access requests, list outstanding requests, approve or deny them, and delete pending requests when they are no longer needed.

  5. Mail Server Management – Administrators can now create, list, retrieve, update, and delete mail server configurations used for platform notifications and related email workflows.

  6. User Agreements – Versioned user-agreement management has been added so operators can create agreements, list versions, retrieve a specific version, and fetch the latest published agreement.

  7. Notifications and Alert Preferences – Users can list and update notification preferences, mark notification events read or unread, delete preferences, retrieve their current alert preferences, and subscribe or unsubscribe from email alerts. Administrators can additionally read and update system-wide alert preferences through dedicated Users APIs.

  8. Proxy Instance Patch Management – Operations teams can inspect Squid proxy patch compliance, list available patches, and trigger patch operations from the SDK, aligning infrastructure maintenance with the rest of the administrative surface.

  9. SQS Queue Administration – Backend queue visibility and control have been added. Administrators can list queues, retrieve queue details, and purge a queue when operational cleanup is required.

  10. Service Limits and Quota Alarms – Service quota inspection and alarm configuration are now first-class SDK methods. Operators can retrieve current limits and usage and set quota alarm thresholds before capacity is exhausted.

  11. File Data Load Limits – Concurrent data-load slot limits are now manageable from the SDK through a dedicated Data Load Limits API. Operators can retrieve the configured slot limits per target (S3, S3Athena, DynamoDB, Lake Formation, and Redshift where a data warehouse is configured) and update them. Reading limits requires the system-settings.view permission and updating them requires system-settings.manage; Redshift is omitted when DWH is not configured.

  12. Data Warehouse Cluster Operations – DWH administration has been expanded beyond read-only cluster details. Users can create, update, and delete tenants and WLM queues, submit DWH queries, perform cluster management actions, retrieve operation history, and inspect DWH, OpenSearch, or RAG cluster details.

  13. Platform Configuration – Centralized administration APIs now cover named system configuration, IP allowlisting, allowed user email domains, and system log retention. Operators can read and update these settings without leaving the SDK.

  14. Operational Observability – Platform health and cost visibility have been extended with health checks, resource count and cost stats, disaster-recovery replication status, account-level Lambda concurrency, and monthly application usage retrieval.

  15. Datasource Operations, Sync Control, and ArcGIS Discovery – Datasource operational APIs have been rebuilt around explicit methods: run a read-only datasource action (connectivity tests, sample data, schema discovery), stop a running sync job, enable or disable EventBridge schedules, list ArcGIS items for ingestion, download JDBC dataflow execution logs, and authenticate AWS Transfer file-server users.

  16. Dataset Multi-File Download and Batch Delete Status – Dataset file handling now supports triggering an asynchronous ZIP download across multiple files, and polling batch dataset-delete jobs for completion status.

  17. ETL Job Execution Re-run – Failed or incomplete root ETL job executions can be re-run from the SDK, reducing the need to reconstruct job submissions after a transient failure.

  18. Healthcare Store Import Jobs – HCLS coverage now includes starting a data import job against a managed store, extending the existing store, metadata, and workflow APIs.

  19. SAML Group Mappings and Pseudo-SSO – Identity integrations have been expanded with list and get methods for SAML group-to-role/tag mappings, plus store and retrieve methods for pseudo-SSO encrypted tokens used by embedded access flows.

Enhancements & API Payload Schema Changes (20)

This release introduces enhancements to existing platform capabilities and to the SDK client itself, driven by expanded payload configurations and a clearer authentication and documentation experience. Several payload models changed shape in ways that require code changes; those carry an explicit Migration: note below.

AI & Machine Learning Enhancements

  • Adaptive Model Invocation – AI model update payloads now accept AdaptiveThinking and Effort so administrators can control extended reasoning behavior on models that support it. These properties are also returned on AIModelDetails and in model listings. The previous ModelParameters field on AIModelUpdateInput is no longer generated, and has likewise been dropped from GuardRailChatRequest. Migration: Move update calls off ModelParameters and onto AdaptiveThinking and Effort.

  • Agent Definition Model – Agent creation and update payloads have been restructured around packaged agents. AgentType, EntryPoint, PythonVersion, DisplayName, and SessionSettings have been added, and ResourceAccessMetadata is now a typed AgentResourceAccessMetadata model. Instruction, Tools, and AgentResponseType have been removed. Migration: CreateAgentRequestBody now requires AgentId, AgentName, Description, ModelId, PythonVersion, EntryPoint, and AgentType. Requests that previously sent Instruction and Tools, which were required in the prior contract, must be rewritten against the new packaged-agent fields.

  • Guardrails with Automated Reasoning – Guard rail create, update, and detail payloads now accept AutomatedReasoningPolicies and a PoliciesConfidenceThreshold, connecting guard rails to the new Automated Reasoning Policy resources. GuardRailUpdateRequest also gained IsCrossRegionEnabled, and guard rail summaries now report GuardRailVersion.

  • Knowledge Base and Project Smart Chunking – KnowledgeBaseCreateRequest and AI Project payloads now accept SmartChunking to control document chunking strategy. The previous EnableSmartOrchestration flag on UpdateProjectRequestBody has been removed. Migration: Replace EnableSmartOrchestration with SmartChunking on project updates.

  • AI Project Request and Response Unions – CreateProjectRequestBody is now a union of CreateAiProjectBody and PinUnpinAiProjectBody, and ListProjectsResponse is a union of AiProjectsPage and AiProjectCategoriesPage. Migration: These generate as union wrapper models rather than flat models; construct the specific variant and read results through the resolved instance.

Data Management & Governance Enhancements

  • Dataset Creation Controls – Dataset metadata used on create now includes SerdeProperties, TableProperties, IngestionType, and CostTags, along with QueryId and FileName where those ingestion paths apply. The update payload (UpdateDatasetMetadata) gained SerdeProperties and TableProperties only. Registered dataset responses now also surface SerdeProperties, TableProperties, DataPermissions, DataProfile, and the EDA and refresh status fields.

  • Dataset Listing Pagination – list_datasets responses now include listing_source and next_cursor, supporting cursor-based pagination on the OpenSearch listing path alongside the existing offset-based DynamoDB path.

  • User Data Permissions Model – UserDataPermissions has been replaced field for field: Columns and FilterType are gone, and DataFilterPermissions and TablePermissions take their place. Migration: Rewrite any code that reads or builds column-level filter permissions against the new two-array shape.

  • Cost Budget Multi-Value Tags – Cost budgets now target multiple tag values. TagValue has been replaced by a TagValues array, and CreateSeparateBudgets and CurrencyUnit have been added. BudgetName, BudgetAmount, BudgetPeriod, ActionConfigurations, TagKey, and TagValues are now explicitly required on creation. Migration: Replace the single TagValue string with a TagValues list.

  • Tag Field Naming Corrections – Two long-standing misspellings were corrected in tag payloads: Tagkey is now TagKey on tag listing and cost-tag-by-name responses, and TagsAttched is now TagsAttached on resource tag responses. Cost-tag list responses renamed the collection field from tags to CostTags. Tag creation and update also accept DisplayName; tag update no longer accepts UsersAttached. Migration: Update any field access that relied on the previous spellings or collection name.

Identity & Access Enhancements

  • SAML Group Mapping Multi-Role and Multi-Tag – SAML group mappings now bind a group to multiple roles and tags. The singular RoleId and Tag fields have been replaced by Roles and Tags arrays across the mapping, update, list, and get payloads. List items now include MappingType (previously only the get-by-id payload reported it), and mapping records now report LastModifiedBy and LastModifiedTime. Migration: Replace RoleId with Roles and Tag with Tags.

  • Access Token Update Payload – AccessTokenPutInput has been narrowed to Description, ExpirationAlertOne, and ExpirationAlertTwo; ExpiresOn, Name, and RoleId are no longer accepted on update. Token creation gained CustomTokenId, and token records now report AssumedBy and TokenVersionId. Migration: Set the expiry, name, and role at creation time; they can no longer be changed through update_access_token.

Data Warehouse & Datasource Operations

  • Datasource Action Dispatch – Read-only datasource diagnostics (test connection, sample data, supported item types, and JDBC schema extract) are now invoked through run_datasource_operation with an explicit action query parameter, replacing the older combined operations endpoint.

  • Cluster Operation Payload Unions – The DWH OperationsResponse model is now a union across the individual cluster result shapes (cluster groups, metrics, schemas, tables, query responses, tenants, WLM configuration, and index history) rather than one flat object carrying every field. The corresponding request models InputBody and InputDetails were restructured to carry tenant, WLM queue, and query properties directly. Migration: Read cluster operation results through the resolved union variant instead of optional attributes on a single response object.

  • Datasource Detail and Dataflow Payloads – Datasource create and update payloads (DatasourceDetails) now carry DatasourceConnectorName and DisplayName and no longer expose a top-level ScheduleConfig (schedule settings remain nested under DatasourceConfig). Dataflow updates renamed ItemConfigs to ItemsConfig and added DataflowDescription, RegisteredDatasets, ReloadOption, and ReloadTables. Execution records were substantially expanded with run status, retry, table-mapping, and statistics fields.

Platform-Wide Enhancements

  • Consistent List Pagination Metadata – List responses across the platform now report count, next_available, and total_count in a uniform way, including AWS roles, dashboards, datalab lifecycle configurations, glossaries, roles, users, and NL2SQL sync jobs.

SDK Client & Documentation

  • SSM Parameter Store Authentication – AWS SSM Parameter Store is now the recommended way to construct AmorphicApiClient. PAT tokens, API Gateway URL, and role ID should be read from SSM at runtime. Environment variables remain supported for local development; Secrets Manager remains supported where rotation policy requires it.

  • Usage Metrics Request Model – send_app_usage now accepts a SendAppUsageRequest body and an x_source argument; both are optional in the generated signature. The generated PostAppUsageBody model and the previous _headers-based X-Source override have been removed. Migration: Replace PostAppUsageBody with SendAppUsageRequest and pass x_source directly instead of overriding headers. The wrapper client already performs this mapping when usage stats are enabled. X-Source remains optional on the Amorphic 3.4 API and is scheduled to become required in 3.5.

  • Request, Response, and Error Documentation – SDK method pages now include request and response examples and documented error codes taken from the Amorphic API definition. Model pages render field descriptions, required flags, and defaults when autodoc-pydantic is installed.

  • User Feedback – A dedicated submit_feedback method has been added so product feedback can be posted through the same Management API used for usage metrics.

Deprecated Features & Removed APIs (2)

To align the generated client with the Amorphic 3.4 API contract, the following methods have been removed in v2.0. Please migrate to the replacements below:

  • Datasource Combined Operations – The legacy datasource_operations method has been removed. Migration: Use run_datasource_operation with the appropriate action (testDatasource, getSampleData, getSupportedItemTypes, getSchemas, getSchemaTables, or getTableMetadata).

  • Datasource Schedule Toggle – The legacy schedule_datasource_operation method has been removed. Migration: Use enable_disable_datasource_schedule to enable or disable EventBridge schedules on supported Amorphic and JDBC datasources.