The Azure SDK team is pleased to announce our August 2026 client library releases.

79 packages released this month.

Stable Packages (40)

  • AI Agent Server - Core

  • AI Projects

  • Device Update

  • Document Translation

  • Resource Management - Alertprocessingrules

  • Resource Management - Automation

  • Resource Management - Azure Stack HCI

  • Resource Management - Computelimit

  • Resource Management - Connected Cache

  • Resource Management - Consumption

  • Resource Management - Container Service

  • Resource Management - Container Service Fleet

  • Resource Management - Data Box

  • Resource Management - Data Protection

  • Resource Management - Discovery

  • Resource Management - Dynatrace

  • Resource Management - Edge Order

  • Resource Management - Elastic

  • Resource Management - Event Hubs

  • Resource Management - Guest Configuration

  • Resource Management - Hybrid Compute

  • Resource Management - Hybrid Kubernetes

  • Resource Management - IoT Operations

  • Resource Management - Marketplace

  • Resource Management - NetApp Files

  • Resource Management - Networkcloud

  • Resource Management - Power BI Dedicated

  • Resource Management - Private DNS

  • Resource Management - Prometheusrulegroups

  • Resource Management - Providerhub

  • Resource Management - Relay

  • Resource Management - Reservations

  • Resource Management - Resource-Deployments

  • Resource Management - Resource-Subscriptions

  • Resource Management - SQL Virtual Machine

  • Resource Management - Service Bus

  • Resource Management - Storage

  • Resource Management - Storage Sync

  • Voice Live

  • unknown

Patch Updates (4)

  • AI Evaluation

  • Cosmos DB

  • Event Grid

  • Key Vault - Secrets

Beta Packages (22)

  • AI Agent Server - Core

  • Azure Content Understanding in Foundry Tools

  • Key Vault - Certificates

  • OpenTelemetry Exporter

  • Resource Management - Cognitive Services

  • Resource Management - Compute Fleet

  • Resource Management - Compute Recommender

  • Resource Management - Compute Schedule

  • Resource Management - Container Service

  • Resource Management - Containerserviceaimanager

  • Resource Management - Datadog

  • Resource Management - Fabric

  • Resource Management - Hardware Security Modules

  • Resource Management - Hybrid Compute

  • Resource Management - IoT Hub

  • Resource Management - NetApp Files

  • Resource Management - SQL

  • Storage - Blobs

  • Storage - Files Data Lake

  • Storage - Files Share

  • Storage - Queues

  • unknown

Release highlights

AI Agent Server - Core 2.0.0b8 Changelog

Features Added

  • Added azure.ai.agentserver.core.storage, the protocol-neutral Foundry durable state-store layer. FoundryStateStore.get_or_create(name, ...) is the primary entry point – an async classmethod that resolves (or creates, on first use) the store in one call. Store-level operations act on the bound store: get() (its descriptor), update(...) (its mutable description / tags), and delete() (the whole store, cascade-deleting every item). Item operations are explicit and consistently named: create_item, set_item, get_item, delete_item, list_keys. Store names are path-encoded with base64url on the wire, store-level item_ttl_seconds is configured once at create, optional user_isolation is declared at store create, and trusted callers may delegate end-user partitioning with user_id (x-ms-user-id). Response bodies (StateStore, StateStoreItem, StateStoreItemRef, DeletedStateStore, DeletedStateStoreItem, StateStoreItemKey) are typed model classes generated from a formal TypeSpec contract, not hand-written dataclasses. See the Durable State Store Guide.

Bugs Fixed

  • Fixed span attribute enrichment under opentelemetry-sdk >= 1.43.0, where span._attributes became a BoundedAttributes (backed by ._dict) that no longer supports item assignment. Enrichment now resolves the backing store so agent identity attributes are written on both older and newer OpenTelemetry SDKs.

AI Agent Server - Core 2.0.0b9 Changelog

Features Added

  • Added Agent Server-managed OTLP/gRPC export when OTEL_EXPORTER_OTLP_PROTOCOL=grpc is configured.
  • Task-record schema cleanup: framework-reserved wire keys in the persisted task record no longer carry a leading _ (e.g. the task_name tag; the schema_version / last_input_id / turn_started_at / retry_attempt / steering payload keys; the input / steering_input_<seq> / output attachment keys) — only the __attachment_ref__ discriminator keeps its marker. The source stamp now includes hosting_environment (from FOUNDRY_HOSTING_ENVIRONMENT), and the payload now carries a schema_version (currently "1"). Tasks persisted before this change (lacking payload.schema_version) are deleted rather than recovered by the recovery scan.
  • Added a resilient task primitive for building long-running agents that survive container restarts, out-of-memory kills, and redeployments. Decorate an async function with @task (one-shot) or @multi_turn_task (multi-turn conversations); the framework persists task state to a task store and automatically recovers and re-invokes in-flight work after a crash. Available from azure.ai.agentserver.core.tasks, with TaskContext, TaskRun, configurable retries (RetryPolicy), cancellation, steering, and a typed exception set (TaskFailed, TaskCancelled, TaskConflictError, TaskManagerNotInitialized, and others). See the Resilient Task Developer Guide.
  • Added an event streaming API (azure.ai.agentserver.core.streaming) for publishing incremental task output to one or more subscribers, with in-memory and file-backed buffering and live or replay delivery. This makes it straightforward to serve Server-Sent Events (SSE) responses that a client can disconnect from and resume. See the Streaming Developer Guide.
  • Exposed resolve_state_subdir(name) on the public azure.ai.agentserver.core surface. It resolves an on-disk state subdirectory (e.g. "tasks", "streams", "responses") under the shared agent-server state root (AGENTSERVER_STATE_ROOT, or ~/.agentserver when unset), so protocol packages persist state under the same operator-controlled root.

AI Agent Server - Core 2.0.0b10 Changelog

Features Added

  • Added public MiddlewareFactory and StreamContent typing aliases for host middleware and streaming helpers.
  • Added set_resilient_tasks_enabled / resilient_tasks_enabled to azure.ai.agentserver.core.tasks — a process-global switch (default off) that force-enables the resilient TaskManager’s startup recovery scan even before any durable task is declared (useful when tasks are registered lazily after startup).

Bugs Fixed

  • AgentServerHost no longer makes a blocking hosted task-store list() round-trip (plus credential-token acquisition) at startup unless resilient tasks are actually in use. The TaskManager is still constructed so get_task_manager() and .run() / .start() keep working, but its network-backed startup recovery scan now runs only when either at least one durable task (@task / @multi_turn_task) has been declared or set_resilient_tasks_enabled(True) was called. Apps that use tasks keep automatic crash recovery with no code change; plain servers (e.g. invocations-only hosts) no longer pay tens of seconds of startup latency for a scan that has nothing to recover.
  • Cleaned up AgentServerHost public signatures so inherited middleware typing does not expose Starlette private type aliases.

AI Agent Server - Core 2.0.0b11 Changelog

Features Added

  • Added a shared experimental decorator for marking Agent Server preview feature surfaces with docstring notes and one-time runtime warnings. The resilient task primitive and Foundry storage public APIs are now marked experimental.

Breaking Changes

  • Removed the unused public AgentServerHost.sse_keepalive_stream helper.

AI Agent Server - Core 2.0.0 Changelog

Other Changes

  • Promoted azure-ai-agentserver-core to GA.

AI Agent Server - Core 2.1.0b1 Changelog

Breaking Changes

  • Removed TaskMetadata, TaskContext.metadata, and TaskRun.metadata. Durable application state now belongs in an explicit FoundryStateStore and no longer shares task lifecycle PATCHes or lease renewal. Typed task inputs may carry a top-level call_id, which the framework restores in FoundryAgentRequestContext for every handler attempt.

Features Added

  • FoundryStateStore now uses a file-backed local fallback when AgentServer is not hosted, and item operations accept an explicit call_id for forwarding recovered durable work outside the original request context.

AI Evaluation 1.18.2 Changelog

Features Added

  • Added support for admin-connected (BYO) judge models in prompty-based evaluators. Models configured with byo_model and project_endpoint are routed through the Foundry project Responses API, which resolves the connection and its authentication.

Bugs Fixed

  • Updated TaskNavigationEfficiencyEvaluator validation errors to name both accepted input aliases: response / actions and ground_truth / expected_actions.

AI Evaluation 1.18.3 Changelog

Features Added

  • Added support for Entra authentication for evaluation results emitted to Application Insights.

AI Projects 2.4.0 Changelog

Breaking Changes

Breaking changes in beta methods:

  • Method .beta.evaluators.create_generation_job renamed to .beta.evaluators.begin_create_generation_job and is now a long-running operation returning LROPoller[EvaluatorVersion] (previously returned EvaluatorGenerationJob).
  • Method .beta.datasets.create_generation_job renamed to .beta.datasets.begin_create_generation_job and is now a long-running operation returning LROPoller[DataGenerationJobResult].
  • Method .beta.agents.create_optimization_job renamed to .beta.agents.begin_create_optimization_job and is now a long-running operation returning LROPoller[OptimizationJobResult].

Features Added

  • New stable toolbox tool ToolSearchToolboxTool (discriminator toolbox_search) for storing a tool-search tool in a toolbox. This replaces ToolboxSearchPreviewToolboxTool, which is still present but will be removed in a future release of the package. Please migrate your code to use the stable tool.
  • New class TaskGenerationDataGenerationJobOptions (discriminator task_generation) with data generation job options for multi-turn evaluation scenarios.
  • Support for non-fatal input-quality advisories from rubric evaluator generation. See new class RubricGenerationInputQualityWarning and new enums RubricGenerationInputQualityWarningCode, RubricGenerationInputQualityWarningSeverity, and RubricGenerationInputQualityWarningSource.
  • New enum GenerationWarningType.
  • New enum AgentIdentityStatus and new optional status property on class AgentIdentity.
  • New read-only property input_quality_warnings on class EvaluatorGenerationJob.
  • New read-only properties generation_job_id and warnings on class EvaluatorVersion.
  • New optional property max_stalls on class OptimizationOptions.

Azure Content Understanding in Foundry Tools 1.2.0b3 Changelog

Features Added

Cosmos DB 4.16.3 Changelog

Bugs Fixed

  • Fixed regression introduced in 4.16.0 on 47105 for complete-partition-key queries scanning documents instead of using partition-key routing, which caused excessive RU consumption and latency for aggregates such as COUNT. See PR 48237

Device Update 1.1.0 Changelog

Features Added

  • Regenerated the client library from the TypeSpec specification for the Device Update for IoT Hub 2026-06-01 GA API version.
  • Added support for configuring TLS-secured update payload downloads through the downloadSecurity property when creating or updating a deployment.

Document Translation 2.0.0 Changelog

Breaking Changes

  • Changed the default service API version from 2024-05-01 to 2026-03-01. To keep the previous behavior, pass api_version="2024-05-01" when constructing the client.
  • Models are no longer re-exported from the top-level azure.ai.translation.document namespace and must now be imported from azure.ai.translation.document.models. This affects TranslationGlossary, TranslationTarget, DocumentTranslationInput, TranslationStatus, DocumentStatus, DocumentTranslationError, DocumentTranslationFileFormat, and StorageInputType. For example, replace from azure.ai.translation.document import TranslationStatus with from azure.ai.translation.document.models import TranslationStatus. The clients (DocumentTranslationClient, SingleDocumentTranslationClient), DocumentTranslationApiVersion, and DocumentTranslationLROPoller remain available at the top level.

Features Added

  • Added support for the 2026-03-01 service API version, which is now the default.
  • Added image translation support: the translate_text_within_image keyword on begin_translation for batch requests, and a translate_text_within_image keyword on SingleDocumentTranslationClient.translate for single document requests. When enabled, each document’s status also reports image scan usage.
  • Added the deployment_name property to TranslationTarget and the deployment_name keyword to begin_translation to specify the deployment name of the custom translation model for a batch translation request.
  • Added the deployment_name keyword to SingleDocumentTranslationClient.translate for single document translation requests.
  • Added the deployment_name property to DocumentStatus, exposing the deployment name of the custom translation model used for the translation.
  • Added image scan reporting to DocumentStatus: image_characters_detected, images_charged, total_image_scans_succeeded, and total_image_scans_failed.
  • Added image scan totals to TranslationStatusSummary: total_image_scans_succeeded, total_image_scans_failed, and total_images_charged.
  • Added the BatchOptions model for specifying batch translation options.

Event Grid 4.22.1 Changelog

Bugs Fixed

  • Fixed a security issue where the legacy EventGridPublisherClient (sync and async) could re-send credential headers to a different host when following an HTTP 3xx redirect. A SensitiveHeaderCleanupPolicy is now added to the pipeline to strip Authorization, aeg-sas-key and aeg-sas-token headers on cross-host redirects.

Key Vault - Certificates 4.12.0b2 Changelog

Bugs Fixed

  • Fixed a replay bug in the challenge authentication policy where a request copy stashed on the shared policy instance was never cleared, allowing one request’s method, URL, and body to leak into a later, unrelated request made by the same client (for example, after an Entra ID Continuous Access Evaluation challenge). The original request is now stored per-request instead of on the policy instance. This mirrors the fix already applied to azure-keyvault-keys and azure-keyvault-administration in #47742.

Key Vault - Secrets 4.11.1 Changelog

Bugs Fixed

  • Fixed a replay bug in the challenge authentication policy where a request copy stashed on the shared policy instance was never cleared, allowing one request’s method, URL, and body to leak into a later, unrelated request made by the same client (for example, after an Entra ID Continuous Access Evaluation challenge). The original request is now stored per-request instead of on the policy instance. This mirrors the fix already applied to azure-keyvault-keys and azure-keyvault-administration in #47742.

OpenTelemetry Exporter 1.0.0b56 Changelog

Features Added

  • Retry payloads for which request is sent successfully but no response is received, per SPEC. (#47870)

Bugs Fixed

  • Propagate main agent attribute to child spans (#47950)
  • Live Metrics now honors the APPLICATIONINSIGHTS_AUTHENTICATION_STRING environment variable for AAD authentication as a fallback when no explicit credential is supplied and local authentication is disabled. (#48284)
  • Fix a memory leak where exporters registered as OneSettings configuration callbacks were retained for the process lifetime; bound-method callbacks are now held via weak references so discarded exporters can be garbage collected. (#48379)

Resource Management - Alertprocessingrules 1.0.0 Changelog

Resource Management - Automation 2.0.0 Changelog

Resource Management - Azure Stack HCI 8.1.0 Changelog

Resource Management - Cognitive Services 15.0.0b4 Changelog

Resource Management - Compute Fleet 2.0.0b3 Changelog

Resource Management - Compute Recommender 1.0.0b2 Changelog

Resource Management - Compute Schedule 1.2.0b2 Changelog

Resource Management - Computelimit 1.3.0 Changelog

Resource Management - Connected Cache 1.0.0 Changelog

Resource Management - Consumption 11.0.0 Changelog

Resource Management - Container Service 41.4.0 Changelog

Resource Management - Container Service 41.5.0 Changelog

Resource Management - Container Service 41.6.0b1 Changelog

Resource Management - Container Service Fleet 4.0.0 Changelog

Resource Management - Containerserviceaimanager 1.0.0b1 Changelog

Resource Management - Data Box 4.0.0 Changelog

Resource Management - Data Protection 2.1.0 Changelog

Resource Management - Datadog 3.0.0b1 Changelog

Resource Management - Discovery 1.0.0 Changelog

Resource Management - Dynatrace 3.0.0 Changelog

Resource Management - Edge Order 3.0.0 Changelog

Resource Management - Elastic 3.0.0 Changelog

Resource Management - Event Hubs 12.0.0 Changelog

Resource Management - Fabric 1.1.0b1 Changelog

Resource Management - Guest Configuration 1.0.0 Changelog

Resource Management - Hardware Security Modules 2.0.0b1 Changelog

Resource Management - Hybrid Compute 9.1.0b5 Changelog

Resource Management - Hybrid Compute 10.0.0 Changelog

Resource Management - Hybrid Kubernetes 2.0.0 Changelog

Resource Management - IoT Hub 5.0.0b4 Changelog

Resource Management - IoT Operations 1.2.0 Changelog

Resource Management - Marketplace 1.0.0 Changelog

Resource Management - NetApp Files 17.1.0 Changelog

Resource Management - NetApp Files 18.0.0b2 Changelog

Resource Management - Networkcloud 3.0.0 Changelog

Resource Management - Power BI Dedicated 2.0.0 Changelog

Resource Management - Private DNS 2.0.0 Changelog

Resource Management - Prometheusrulegroups 1.0.0 Changelog

Resource Management - Providerhub 1.0.0 Changelog

Resource Management - Relay 2.0.0 Changelog

Resource Management - Reservations 3.0.0 Changelog

Resource Management - Resource-Deployments 1.0.0 Changelog

Resource Management - Resource-Subscriptions 1.0.0 Changelog

Resource Management - SQL 5.0.0b1 Changelog

Resource Management - SQL Virtual Machine 1.0.0 Changelog

Resource Management - Service Bus 10.0.0 Changelog

Resource Management - Storage 25.1.0 Changelog

Resource Management - Storage Sync 2.0.0 Changelog

Storage - Blobs 12.31.0b1 Changelog

Features Added

  • Added support for service version 2026-10-06.
  • Added support for parsing List Blobs responses in Apache Arrow format by passing response_format="arrow" keyword to ContainerClient’s list_blobs, walk_blobs, and list_blob_names APIs (response_format defaults to "auto", and "auto" currently means "xml"). The use of "response_format=arrow" requires nanoarrow to be installed. This change also enables the new end_before keyword for range-based listing.
  • Added access tier information to the response of BlobClient’s download_blob API. The blob_tier, blob_tier_inferred, blob_tier_change_time, and smart_access_tier properties are now populated on the downloaded blob’s properties.
  • The service-calculated CRC64 is now surfaced as content_crc64 on the response of BlobClient upload operations (stage_block, stage_block_from_url, upload_page, upload_pages_from_url, append_block, append_block_from_url, upload_blob, and upload_blob_from_url) in addition to content_md5 when a content MD5 is provided with the request.

Storage - Files Data Lake 12.26.0b1 Changelog

Features Added

  • Added support for service version 2026-10-06.

Storage - Files Share 12.27.0b1 Changelog

Bugs Fixed

  • Return type of get_share_access_policies is now the publicly exposed SignedIdentifier model containing the publicly exposed AccessPolicy model. Previously we were leaking the internal generated SignedIdentifier model without exposing it in the api.

Features Added

  • Added support for service version 2026-10-06.
  • Added list_ranges and list_ranges_diff APIs to ShareFileClient which return an auto-paging iterable of FileRange and support continuation tokens and results_per_page for enumerating file ranges across multiple service responses.

Storage - Queues 12.18.0b1 Changelog

Bugs Fixed

  • Return type of get_access_policies is now the publicly exposed AccessPolicy model. Previously we were leaking the internal generated AccessPolicy model without exposing it in the api.

Features Added

  • Added support for service version 2026-10-06.

Voice Live 1.3.0 Changelog

Bugs Fixed

  • Image Input Field Rename: Renamed RequestImageContentPart.url to image_url. Update image input construction to use image_url= instead of url=.

Breaking Changes

  • Default API Version Update: Changed the SDK default API version from 2026-06-01-preview to the GA version 2026-07-15. Pass api_version="2026-06-01-preview" explicitly to keep the previous default behavior.

Features Added

  • Azure Realtime Native Voice Support: Added AzureRealtimeNativeVoice and AzureRealtimeNativeVoiceName, and expanded voice fields to accept Azure realtime native voices.
  • Input Text Streaming Support: Added ClientEventInputTextDelta and ClientEventInputTextDone for incrementally streaming text input into existing conversation items.
  • Hosted Agent Invocation Input: Added invoke_input to ResponseCreateParams and ServerEventResponseInvocationDelta for hosted agent invocation passthrough data.
  • Echo Cancellation Configuration: Added EchoCancellationReferenceSource and new reference_source / channels options on AudioEchoCancellation to support both the default server loopback reference path and client-provided stereo echo reference input.
  • Parallel Tool Call Control: Added parallel_tool_calls to session models so callers can control whether tool calls may run in parallel.
  • Session Expiration: Added expires_at to ResponseSession, a server-set datetime indicating when the session expires.

unknown 1.0.0b9 Changelog

Other Changes

  • Raised the azure-ai-agentserver-core dependency floor to >=2.0.0b8.

unknown 1.0.0b1 Changelog

unknown 1.0.0b2 Changelog

unknown 1.0.0b1 Changelog

unknown 2.0.0b0 Changelog

Breaking Changes

  • Removed a-prefixed async convenience generator methods from the sync ResponseEventStream and sync builder classes. Use azure.ai.agentserver.responses.aio.ResponseEventStream for async streaming convenience methods.
  • Replaced generated model classes in azure.ai.agentserver.responses.models with dict-native TypedDict contracts. Model constructors such as ItemMessage(...) and CreateResponse(...) now produce plain dictionaries instead of generated model instances.
  • Removed runtime model-class behavior from response protocol models. Code should no longer rely on attribute access, isinstance(..., ModelType), .as_dict(), or generated model base-class behavior.

  • Replaced most generated enum classes with string literal type aliases. Use string values directly for protocol fields, for example "completed", "message", or "function_call_output".

  • The resilient-task input persisted for a store=true background response now carries a single user_id_key (the durable per-user partition key) instead of the previous user_isolation_key / chat_isolation_key pair; conversation scoping continues to use conversation_id.

Bugs Fixed

  • Steering now works on the first turn of a conversation. In a steerable_conversations=true deployment, the first turn (a request with no conversation_id and no previous_response_id) is now hosted on the multi-turn chain primitive instead of a one-shot task. Because all turns of a chain share a stable conversation_chain_id — and therefore the same backing task — a steered turn posted onto an in-flight first turn previously queued onto a one-shot task that completed and auto-deleted before draining the queued input, leaving the steered turn stuck in_progress. The first turn now suspends between turns and drains queued steering inputs correctly.

  • context.conversation_chain_id is now stable across every turn of a conversation. Previously it returned the raw previous_response_id (the immediate predecessor), so it shifted on every turn after the second — breaking its use as a stable per-conversation key (e.g. an upstream SDK session id). It is now derived from the partition key embedded in the chain’s response IDs (which every turn shares), so all turns of a chain — and the resilient task that backs them — resolve to the same identity. The value is now an opaque, agent/session-scoped hash rather than a raw id. (Known limitation: a client that supplies its own response_id with a mismatched embedded partition can shift the chain identity for later turns.)

Features Added

  • Added the azure.ai.agentserver.responses.aio namespace with async ResponseEventStream convenience generators that use the same method names as the sync stream, such as output_item_message() and output_item_compaction().
  • Added local TypedDict model contract generation for the Responses protocol, including generated type aliases, union aliases, and py.typed packaging support.

  • Added dict-native wire payload helpers and request validators for validating protocol payloads without depending on generated model internals.

  • ResponseContext.conversation_chain_id (and the resilient task id it backs) now follows the native id convention: cchain_<partition><scope> for conversation-scoped chains, rchain_<partition><scope> for steerable response-linkage chains, or the response_id verbatim for a non-steerable one-shot. The id embeds the chain’s partition key for co-location and carries a deterministic (agent, session) scope; task_id == conversation_chain_id exactly. Replaces the previous opaque resilient-resp-<32-hex> form.

  • Resilient background responses. ResponsesServerOptions(resilient_background=True) makes store=true, background=true responses survive process crashes: the framework persists handler progress and re-invokes the registered handler on the next process start when a prior attempt did not reach a terminal event. Defaults to False.

  • Steerable conversations. ResponsesServerOptions(steerable_conversations=True) lets clients post a new turn on an in-flight conversation; the running handler is woken (via the cancellation signal, distinguished by context.pending_input_count > 0), drains the queued input on a fresh invocation, and the turns are linked in a stable conversation chain. Defaults to False.

  • ResponseContext resilience + steering surface. Flat fields stamped on each invocation: context.is_recovery, context.is_steered_turn, context.pending_input_count, and context.conversation_chain_id (a stable identifier shared by every turn of a conversation chain, usable as a key into application-side session state).

  • Developer checkpoints. yield stream.checkpoint() persists the current response snapshot at a developer-chosen boundary (gated to resilient background responses; a no-op otherwise; backpressured and idempotent). On a recovered entry, context.persisted_response exposes the last persisted snapshot so the handler can seed its stream and resume — the basis of the one-OutputItem-per-phase recovery pattern.

  • internal_metadata. A single-turn, platform-internal MutableMapping[str, Any] on output items (item.internal_metadata) and on the response (stream.internal_metadata). It is persisted with the response (so it is available on recovery) and is always stripped before any client-facing HTTP/SSE payload, and on ingress. Distinct from the public ResponseObject.metadata.

  • context.conversation_chain_metadata. Cross-turn, named-scope, explicit-flush() resilient metadata over a conversation chain, typed by the public ConversationChainMetadataNamespace Protocol.

  • await context.exit_for_recovery(). A single uniform graceful-shutdown recovery primitive that works in every handler shape (coroutine, async generator, sync) — it raises ResponseExitForRecovery internally to leave the response in_progress for next-lifetime recovery.

  • Stream recovery. SSE events are persisted incrementally; clients reconnect with GET /responses/{id}?stream=true&starting_after=<event_id> and resume from their last received event.

  • Response acceptor hook. Register @app.response_acceptor to customize the response shape returned when a turn is queued behind an active steerable conversation.

  • Storage. FileResponseStore is exported from azure.ai.agentserver.responses and is the default local-development store (under ${AGENTSERVER_STATE_ROOT:-~/.agentserver}/responses/) when no store= is supplied in a non-hosted environment; pass store=InMemoryResponseProvider() to opt out. The AGENTSERVER_STATE_ROOT environment variable sets the local state storage root. A typed ResponseAlreadyExistsError is raised by the response-store providers on a duplicate create_response (the idempotent-create signal on recovery).

  • Handlers are async def. @app.response_handler requires an async handler with the (request, context, cancellation_signal) signature so it can observe the asyncio.Event cancellation signal.

unknown 1.0.0 Changelog

Features Added

  • Workspace:
  • New paged list response models: PagedConversation, PagedInvestigation.
  • New StorageMountProtocol enum for controlling storage mount protocols.
  • New tools.cancel_run_lro long-running cancellation flow (in addition to the existing immediate cancel_run).
  • Investigation update is now exposed via a documented sample.
  • Bookshelf: knowledge-base surface is significantly redesigned around a single KnowledgeBasesOperations group that exposes the full lifecycle in one place:
  • Lifecycle: create_or_update, get, delete, plus get_operation_status for polling long-running operations.
  • Indexing: start_indexing and cancel_indexing, with results modeled via KnowledgeBaseIndexingOperationResponse, IndexingOperationResult, IndexingMetrics, and LastIndexingRun.
  • Search: new search operation taking SearchRequest and returning SearchResponse, including SearchResultItem with Citation and CitationType for citation-aware results.
  • LRO results: status responses now use KnowledgeBaseOperationResponse / KnowledgeBaseSearchOperationResponse; create/update returns the KnowledgeBase resource.
  • New enum KnowledgeBaseOperationType.

Breaking Changes

Note: these are breaking changes only relative to the 1.0.0b1 preview release. As a first stable (GA) release, 1.0.0 is the new compatibility baseline going forward.

  • Bookshelf: the KnowledgeBaseVersionsOperations operation group is removed. Knowledge-base versioning has been folded into the unified KnowledgeBasesOperations group; callers using client.knowledge_base_versions.<method> must migrate to the equivalent method on client.knowledge_bases.
  • Bookshelf: the models KnowledgeBaseOperationStatus and KnowledgeBaseVersion are removed. Operation-status payloads are now typed as KnowledgeBaseOperationResponse, KnowledgeBaseIndexingOperationResponse, or KnowledgeBaseSearchOperationResponse depending on the operation; create/update returns the KnowledgeBase resource.
  • The preview API version 2026-02-01-preview is no longer listed as a supported value for the api_version kwarg. Both WorkspaceClient and BookshelfClient now default to 2026-06-01. Pinning to the removed preview value is not supported in the GA SDK.
  • Workspace: the investigations long-running operation status model, previously generated as ResourceOperationStatusInvestigationInvestigationError, has been renamed to InvestigationOperationStatus. The payload is unchanged; only the model name differs.

unknown 2.0.0b1 Changelog

Other Changes

  • Cleaned up the public API surface by moving validation-only error helpers to a private implementation module and renaming runtime terminal/replay helpers as private.
  • Bumped the minimum azure-ai-agentserver-core dependency to >=2.0.0b10, which adds an opt-in gate for resilient-task startup recovery. The resilient Responses samples now call set_resilient_tasks_enabled(True) to explicitly opt in, mirroring the invocations resilient samples.

unknown 2.0.0 Changelog

Features Added

  • First stable release of the Azure AI Agent Server Responses client library.

Breaking Changes

  • Removed the duplicate azure.ai.agentserver.responses.get_input_expanded export. Import it from azure.ai.agentserver.responses.models instead.

unknown 2.1.0b1 Changelog

Breaking Changes

  • Removed ResponseContext.conversation_chain_metadata and the ConversationChainMetadataNamespace protocol. Resilient response applications now persist cross-turn state explicitly with FoundryStateStore.

Latest Releases

View all the latest versions of Python packages here.

Installation Instructions

To install the latest beta version of the packages, copy and paste the following commands into a terminal:

$> pip install azure-ai-agentserver-core==2.0.0b8
$> pip install azure-ai-agentserver-core==2.0.0b9
$> pip install azure-ai-agentserver-core==2.0.0b10
$> pip install azure-ai-agentserver-core==2.0.0b11
$> pip install azure-ai-agentserver-core==2.0.0
$> pip install azure-ai-agentserver-core==2.1.0b1
$> pip install azure-ai-agentserver-responses==1.0.0b9
$> pip install azure-ai-agentserver-responses==2.0.0b0
$> pip install azure-ai-agentserver-responses==2.0.0b1
$> pip install azure-ai-agentserver-responses==2.0.0
$> pip install azure-ai-agentserver-responses==2.1.0b1
$> pip install azure-ai-contentunderstanding==1.2.0b3
$> pip install azure-ai-discovery==1.0.0
$> pip install azure-ai-evaluation==1.18.2
$> pip install azure-ai-evaluation==1.18.3
$> pip install azure-ai-projects==2.4.0
$> pip install azure-ai-translation-document==2.0.0
$> pip install azure-ai-voicelive==1.3.0
$> pip install azure-cosmos==4.16.3
$> pip install azure-eventgrid==4.22.1
$> pip install azure-iot-deviceupdate==1.1.0
$> pip install azure-keyvault-certificates==4.12.0b2
$> pip install azure-keyvault-secrets==4.11.1
$> pip install azure-mgmt-alertprocessingrules==1.0.0
$> pip install azure-mgmt-automation==2.0.0
$> pip install azure-mgmt-azurestackhci==8.1.0
$> pip install azure-mgmt-cognitiveservices==15.0.0b4
$> pip install azure-mgmt-compute-bulkaction==1.0.0b1
$> pip install azure-mgmt-compute-bulkaction==1.0.0b2
$> pip install azure-mgmt-computefleet==2.0.0b3
$> pip install azure-mgmt-computelimit==1.3.0
$> pip install azure-mgmt-computerecommender==1.0.0b2
$> pip install azure-mgmt-computeschedule==1.2.0b2
$> pip install azure-mgmt-connectedcache==1.0.0
$> pip install azure-mgmt-consumption==11.0.0
$> pip install azure-mgmt-containerservice==41.4.0
$> pip install azure-mgmt-containerservice==41.5.0
$> pip install azure-mgmt-containerservice==41.6.0b1
$> pip install azure-mgmt-containerserviceaimanager==1.0.0b1
$> pip install azure-mgmt-containerservicefleet==4.0.0
$> pip install azure-mgmt-containerservicepreparedimgspec==1.0.0b1
$> pip install azure-mgmt-databox==4.0.0
$> pip install azure-mgmt-datadog==3.0.0b1
$> pip install azure-mgmt-dataprotection==2.1.0
$> pip install azure-mgmt-discovery==1.0.0
$> pip install azure-mgmt-dynatrace==3.0.0
$> pip install azure-mgmt-edgeorder==3.0.0
$> pip install azure-mgmt-elastic==3.0.0
$> pip install azure-mgmt-eventhub==12.0.0
$> pip install azure-mgmt-fabric==1.1.0b1
$> pip install azure-mgmt-guestconfig==1.0.0
$> pip install azure-mgmt-hardwaresecuritymodules==2.0.0b1
$> pip install azure-mgmt-hybridcompute==9.1.0b5
$> pip install azure-mgmt-hybridcompute==10.0.0
$> pip install azure-mgmt-hybridkubernetes==2.0.0
$> pip install azure-mgmt-iothub==5.0.0b4
$> pip install azure-mgmt-iotoperations==1.2.0
$> pip install azure-mgmt-marketplace==1.0.0
$> pip install azure-mgmt-netapp==17.1.0
$> pip install azure-mgmt-netapp==18.0.0b2
$> pip install azure-mgmt-networkcloud==3.0.0
$> pip install azure-mgmt-powerbidedicated==2.0.0
$> pip install azure-mgmt-privatedns==2.0.0
$> pip install azure-mgmt-prometheusrulegroups==1.0.0
$> pip install azure-mgmt-providerhub==1.0.0
$> pip install azure-mgmt-relay==2.0.0
$> pip install azure-mgmt-reservations==3.0.0
$> pip install azure-mgmt-resource-deployments==1.0.0
$> pip install azure-mgmt-resource-subscriptions==1.0.0
$> pip install azure-mgmt-servicebus==10.0.0
$> pip install azure-mgmt-sql==5.0.0b1
$> pip install azure-mgmt-sqlvirtualmachine==1.0.0
$> pip install azure-mgmt-storage==25.1.0
$> pip install azure-mgmt-storagesync==2.0.0
$> pip install azure-monitor-opentelemetry-exporter==1.0.0b56
$> pip install azure-storage-blob==12.31.0b1
$> pip install azure-storage-file-datalake==12.26.0b1
$> pip install azure-storage-file-share==12.27.0b1
$> pip install azure-storage-queue==12.18.0b1

Feedback

If you have a bug or feature request for one of the libraries, please post an issue to GitHub.