[agentserver] azure-ai-agentserver-responses package#46052
Open
[agentserver] azure-ai-agentserver-responses package#46052
Conversation
- added type spec model generation - add model validator generation - creating a server
* trying * generate contract models * add validator generator * fix model generation * add more unit tests * fix conflict * refined model generation
* trying * generate contract models * add validator generator * fix model generation * add more unit tests * fix conflict * refined model generation * renamed the pacakge
* create response * cancel and delete * fix options
The latest opentelemetry-sdk added _on_ending to the SpanProcessor interface, causing AttributeError at runtime. Add the required no-op method to fix compatibility. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…_id resolution and stamping) Agent-Logs-Url: https://github.com/Azure/azure-sdk-for-python/sessions/40114cbe-e583-4912-abc7-75c551a7e12c Co-authored-by: ankitbko <3169316+ankitbko@users.noreply.github.com>
…ID Resolution) Agent-Logs-Url: https://github.com/Azure/azure-sdk-for-python/sessions/40114cbe-e583-4912-abc7-75c551a7e12c Co-authored-by: ankitbko <3169316+ankitbko@users.noreply.github.com>
… private imports, simplify model access in samples Agent-Logs-Url: https://github.com/Azure/azure-sdk-for-python/sessions/40114cbe-e583-4912-abc7-75c551a7e12c Co-authored-by: ankitbko <3169316+ankitbko@users.noreply.github.com>
…n hint priority, add invalid header test Agent-Logs-Url: https://github.com/Azure/azure-sdk-for-python/sessions/40114cbe-e583-4912-abc7-75c551a7e12c Co-authored-by: ankitbko <3169316+ankitbko@users.noreply.github.com>
* update otel * refined readme * updated model validator * fix model
…cing dupe protection Thread 17: Replace ErrorResponse.create() with create_error_response() and AgentLogger.get() with get_logger() module-level functions. Updated all call sites across core, invocations, and responses packages. Thread 18: Added Breaking Changes section to CHANGELOG.md documenting the API renames, removed constants, and health endpoint rename. Thread 21: Added module-level sentinel flags to _setup_trace_export, _setup_log_export, _setup_otlp_trace_export, and _setup_otlp_log_export to prevent duplicate exporters across multiple TracingHelper instantiations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ider Migrated FoundryStorageProvider from raw httpx.AsyncClient to azure.core.AsyncPipelineClient, gaining built-in retry, logging, distributed tracing, and bearer-token authentication via AsyncBearerTokenCredentialPolicy. - Removed manual _auth_headers() — handled by pipeline policy - Removed http_client constructor parameter — pipeline is internal - Updated _foundry_errors.py to use azure.core.rest.HttpResponse - Removed httpx from pyproject.toml dependencies - Updated all unit tests to mock send_request instead of httpx methods Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
CreateResponse always has a model attribute, so the defensive getattr is unnecessary boilerplate. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Responses package has been split into a separate PR on the agentserver/responses branch for independent review and iteration. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This reverts commit 42e88b0.
…langgraph from main These packages were inadvertently removed in an earlier commit. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ly (#46069) Add set_current_span/detach_context to TracingHelper so the invoke_agent span is attached as the active OTel context span. This ensures child spans created by framework handlers are parented under invoke_agent instead of appearing as siblings. - TracingHelper.set_current_span() attaches the span to context - TracingHelper.detach_context() restores the previous context - InvocationHandler propagates span_token through both streaming and non-streaming cleanup paths - Added test_span_parenting.py to verify parent-child relationship Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: set OTel span as current context for proper parent-child span relationships
The responses handler creates the OTel span via start_request_span() which
uses start_span() internally - a detached span not set as the active context.
This causes child spans from user handler code (e.g. Agent Framework) to appear
as separate root spans instead of children.
Use set_current_span() after creating the span so downstream code sees it as
the active parent. Detach the context when streaming completes or in the
finally block for non-streaming paths.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: rename span operation from create_response to invoke_agent
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: use agent name in span name instead of model name
The span name is now 'invoke_agent {name}:{version}' matching the
pattern used by Microsoft Agent Framework.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
Pull request overview
This PR introduces the new azure-ai-agentserver-responses package (split out for independent review) and concurrently refactors packaging/hosting infrastructure across the agentserver repo (core + invocations), including significant removals of the previous langgraph/agentframework adapters and samples.
Changes:
- Added Responses server hosting primitives (execution context, replayable SSE event subject,
ResponseContext, and server options) plus public exports and packaging metadata. - Refactored
azure-ai-agentserver-coreand introducedazure-ai-agentserver-invocationsprotocol package with updated packaging metadata, new tests, and new samples. - Removed the LangGraph and Agent Framework adapter packages’ code/tests/samples and associated metadata files from the repo.
Reviewed changes
Copilot reviewed 196 out of 325 changed files in this pull request and generated 10 comments.
Show a summary per file
| File | Description |
|---|---|
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/hosting/_execution_context.py | New per-request execution context object for Responses orchestration. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/hosting/_event_subject.py | New in-process replay subject for SSE event broadcasting. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/hosting/init.py | Public exports for hosting/routing/validation helpers. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/_version.py | Defines the package version constant. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/_response_context.py | Adds ResponseContext runtime object for handlers + history/input access. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/_options.py | Adds strongly-typed server options + env parsing/validation. |
| sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/init.py | Public API surface exports for responses package. |
| sdk/agentserver/azure-ai-agentserver-responses/MANIFEST.in | Ensures py.typed is packaged for responses. |
| sdk/agentserver/azure-ai-agentserver-responses/CHANGELOG.md | Adds responses package changelog entry for initial beta. |
| sdk/agentserver/azure-ai-agentserver-langgraph/tests/unit_tests/test_langgraph_request_converter.py | Removed LangGraph adapter unit tests (package removal). |
| sdk/agentserver/azure-ai-agentserver-langgraph/tests/conftest.py | Removed LangGraph adapter test harness (package removal). |
| sdk/agentserver/azure-ai-agentserver-langgraph/tests/init.py | Removed LangGraph adapter tests package marker. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_react_agent/requirements.txt | Removed LangGraph sample dependency list. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_react_agent/main.py | Removed LangGraph sample agent entrypoint. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_react_agent/.env-template | Removed LangGraph sample env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_agent_with_redis_checkpointer/requirements.txt | Removed LangGraph Redis sample dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_agent_with_redis_checkpointer/main.py | Removed LangGraph Redis sample entrypoint. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_agent_with_redis_checkpointer/README.md | Removed LangGraph Redis sample docs. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/simple_agent_with_redis_checkpointer/.env-template | Removed LangGraph Redis env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_simple/requirements.txt | Removed LangGraph MCP sample dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_simple/mcp_simple.py | Removed LangGraph MCP sample script. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_simple/README.md | Removed LangGraph MCP sample docs. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_simple/.env-template | Removed LangGraph MCP env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_apikey/requirements.txt | Removed LangGraph MCP API-key sample dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_apikey/mcp_apikey.py | Removed LangGraph MCP API-key sample script. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_apikey/README.md | Removed LangGraph MCP API-key sample docs. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/mcp_apikey/.env-template | Removed LangGraph MCP API-key env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/custom_state/requirements.txt | Removed LangGraph custom state sample dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/custom_state/README.md | Removed LangGraph custom state sample docs. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/custom_state/.env-template | Removed LangGraph custom state env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/workflow.py | Removed LangGraph agentic RAG sample workflow. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/tools/retriever_tool.py | Removed LangGraph agentic RAG retriever tool. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/requirements.txt | Removed LangGraph agentic RAG dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/nodes/rewrite_question.py | Removed LangGraph agentic RAG node. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/nodes/generate_query_or_respond.py | Removed LangGraph agentic RAG node. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/nodes/generate_answer.py | Removed LangGraph agentic RAG node. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/edges/grade_documents.py | Removed LangGraph agentic RAG edge logic. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agentic_rag/.env-template | Removed LangGraph agentic RAG env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agent_calculator/requirements.txt | Removed LangGraph calculator sample dependencies. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agent_calculator/langgraph_agent_calculator.py | Removed LangGraph calculator sample implementation. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agent_calculator/README.md | Removed LangGraph calculator sample docs. |
| sdk/agentserver/azure-ai-agentserver-langgraph/samples/agent_calculator/.env-template | Removed LangGraph calculator env template. |
| sdk/agentserver/azure-ai-agentserver-langgraph/mypy.ini | Removed LangGraph package mypy config. |
| sdk/agentserver/azure-ai-agentserver-langgraph/cspell.json | Removed LangGraph package spell config. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/utils.py | Removed LangGraph adapter model utilities. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/response_stream_event_generator.py | Removed LangGraph stream event generator impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/response_output_text_event_generator.py | Removed LangGraph output text generator impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/response_function_call_argument_event_generator.py | Removed LangGraph function-call generator impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/response_event_generator.py | Removed LangGraph generator base classes. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/response_content_part_event_generator.py | Removed LangGraph content part generator impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/item_resource_helpers.py | Removed LangGraph item resource helper impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/item_content_helpers.py | Removed LangGraph item content helper impl. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/response_event_generators/init.py | Removed LangGraph event generator exports. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/langgraph_stream_response_converter.py | Removed LangGraph stream response converter. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/langgraph_state_converter.py | Removed LangGraph state converter interfaces. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/langgraph_response_converter.py | Removed LangGraph non-stream response converter. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/langgraph_request_converter.py | Removed LangGraph request converter. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/models/init.py | Removed LangGraph model exports. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/langgraph.py | Removed LangGraph adapter runtime. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/agentserver/langgraph/init.py | Removed LangGraph package init/export surface. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/ai/init.py | Removed LangGraph namespace marker. |
| sdk/agentserver/azure-ai-agentserver-langgraph/azure/init.py | Removed LangGraph namespace marker. |
| sdk/agentserver/azure-ai-agentserver-langgraph/README.md | Removed LangGraph package readme. |
| sdk/agentserver/azure-ai-agentserver-langgraph/CHANGELOG.md | Removed LangGraph package changelog. |
| sdk/agentserver/azure-ai-agentserver-invocations/tests/test_session_id.py | New tests for session ID resolution + headers. |
| sdk/agentserver/azure-ai-agentserver-invocations/tests/test_server_routes.py | New tests for route registration + OpenAPI route behavior. |
| sdk/agentserver/azure-ai-agentserver-invocations/tests/test_request_limits.py | New tests ensuring request timeout feature is removed. |
| sdk/agentserver/azure-ai-agentserver-invocations/tests/test_invoke.py | New tests for invoke dispatch, streaming, headers, errors. |
| sdk/agentserver/azure-ai-agentserver-invocations/tests/test_get_cancel.py | New tests for GET/cancel flows and error behavior. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/streaming_invoke_agent/streaming_invoke_agent.py | Added SSE streaming sample agent. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/streaming_invoke_agent/requirements.txt | Added minimal requirements for streaming invocation sample. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/simple_invoke_agent/simple_invoke_agent.py | Added minimal JSON invoke sample agent. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/simple_invoke_agent/requirements.txt | Added minimal requirements for simple invoke sample. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/multiturn_invoke_agent/requirements.txt | Added requirements for multi-turn invoke sample. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/multiturn_invoke_agent/multiturn_invoke_agent.py | Added multi-turn session demo sample agent. |
| sdk/agentserver/azure-ai-agentserver-invocations/samples/async_invoke_agent/requirements.txt | Added requirements for async invoke sample. |
| sdk/agentserver/azure-ai-agentserver-invocations/pyrightconfig.json | Updated pyright settings and exclusions. |
| sdk/agentserver/azure-ai-agentserver-invocations/pyproject.toml | Renamed/re-scoped the project to invocations + updated build tooling flags. |
| sdk/agentserver/azure-ai-agentserver-invocations/dev_requirements.txt | Added dev requirements for invocations package testing/tooling. |
| sdk/agentserver/azure-ai-agentserver-invocations/cspell.json | Added spell checker config for invocations package. |
| sdk/agentserver/azure-ai-agentserver-invocations/azure/ai/agentserver/invocations/_version.py | Added invocations version constant. |
| sdk/agentserver/azure-ai-agentserver-invocations/azure/ai/agentserver/invocations/_constants.py | Added protocol constants for invocations. |
| sdk/agentserver/azure-ai-agentserver-invocations/azure/ai/agentserver/invocations/init.py | Added invocations public package surface and docs. |
| sdk/agentserver/azure-ai-agentserver-invocations/azure/ai/agentserver/init.py | Adjusted namespace path init. |
| sdk/agentserver/azure-ai-agentserver-invocations/MANIFEST.in | Updated typed marker inclusion for invocations package. |
| sdk/agentserver/azure-ai-agentserver-invocations/LICENSE | Normalized license text formatting. |
| sdk/agentserver/azure-ai-agentserver-invocations/CHANGELOG.md | Added invocations package changelog for initial beta. |
| sdk/agentserver/azure-ai-agentserver-core/tests/test_server_routes.py | Added core tests for port resolution and 404 behavior. |
| sdk/agentserver/azure-ai-agentserver-core/tests/test_logger.py | Added tests ensuring logger behavior is stable. |
| sdk/agentserver/azure-ai-agentserver-core/tests/test_health.py | Added tests for readiness endpoint behavior. |
| sdk/agentserver/azure-ai-agentserver-core/tests/test_edge_cases.py | Added tests for log level behaviors and edge cases. |
| sdk/agentserver/azure-ai-agentserver-core/tests/env-template | Removed legacy env template. |
| sdk/agentserver/azure-ai-agentserver-core/samples/simple_mock_agent/requirements.txt | Removed legacy sample requirements. |
| sdk/agentserver/azure-ai-agentserver-core/samples/simple_mock_agent/custom_mock_agent_test.py | Removed legacy sample code. |
| sdk/agentserver/azure-ai-agentserver-core/samples/selfhosted_invocation/selfhosted_invocation.py | Added “tier 3” self-hosting invocation sample. |
| sdk/agentserver/azure-ai-agentserver-core/samples/selfhosted_invocation/requirements.txt | Added requirements for self-hosted invocation sample. |
| sdk/agentserver/azure-ai-agentserver-core/samples/mcp_simple/requirements.txt | Removed legacy MCP sample requirements. |
| sdk/agentserver/azure-ai-agentserver-core/samples/bilingual_weekend_planner/run.sh | Removed legacy runner script. |
| sdk/agentserver/azure-ai-agentserver-core/samples/bilingual_weekend_planner/requirements.txt | Removed legacy weekend planner deps. |
| sdk/agentserver/azure-ai-agentserver-core/samples/bilingual_weekend_planner/README.md | Removed legacy weekend planner docs. |
| sdk/agentserver/azure-ai-agentserver-core/samples/bilingual_weekend_planner/.env.sample | Removed legacy env sample. |
| sdk/agentserver/azure-ai-agentserver-core/pyrightconfig.json | Updated pyright exclusions. |
| sdk/agentserver/azure-ai-agentserver-core/pyproject.toml | Updated project metadata, deps, and build tool flags. |
| sdk/agentserver/azure-ai-agentserver-core/doc/azure.ai.agentserver.core.server.rst | Removed legacy Sphinx docs. |
| sdk/agentserver/azure-ai-agentserver-core/doc/azure.ai.agentserver.core.server.common.rst | Removed legacy Sphinx docs. |
| sdk/agentserver/azure-ai-agentserver-core/doc/azure.ai.agentserver.core.server.common.id_generator.rst | Removed legacy Sphinx docs. |
| sdk/agentserver/azure-ai-agentserver-core/doc/azure.ai.agentserver.core.rst | Removed legacy Sphinx docs. |
| sdk/agentserver/azure-ai-agentserver-core/dev_requirements.txt | Updated dev requirements to align with new test stack. |
| sdk/agentserver/azure-ai-agentserver-core/cspell.json | Updated spell checker config for core package. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/server/common/id_generator/id_generator.py | Removed legacy ID generator interface. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/server/common/id_generator/foundry_id_generator.py | Removed legacy Foundry ID generator implementation. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/server/common/agent_run_context.py | Removed legacy AgentRunContext implementation. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/models/projects/_patch_evaluations.py | Removed legacy generated patch customizations. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/models/projects/_patch.py | Removed legacy generated patch entrypoint. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/models/openai/init.py | Removed OpenAI SDK re-export module. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/models/_create_response.py | Removed legacy CreateResponse TypedDict. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/models/init.py | Removed legacy models package exports. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/logger.py | Removed legacy logging/configure module. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/constants.py | Removed legacy constants module. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/_version.py | Updated core version to 2.0.0b1. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/_logger.py | Added lightweight logger facade get_logger(). |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/_errors.py | Added standardized error response envelope builder. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/_constants.py | Added new constants enum for env vars/defaults. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/core/init.py | Reworked core public API surface and removed implicit logging side-effects. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/agentserver/init.py | Adjusted namespace path init. |
| sdk/agentserver/azure-ai-agentserver-core/azure/ai/init.py | Adjusted namespace path init. |
| sdk/agentserver/azure-ai-agentserver-core/azure/init.py | Adjusted namespace path init. |
| sdk/agentserver/azure-ai-agentserver-core/MANIFEST.in | Removed doc packaging include; kept samples/tests. |
| sdk/agentserver/azure-ai-agentserver-core/LICENSE | Normalized license text formatting. |
| sdk/agentserver/azure-ai-agentserver-core/CHANGELOG.md | Added core 2.0.0b1 changelog with features + breaking changes. |
| sdk/agentserver/azure-ai-agentserver-agentframework/tests/unit_tests/test_agent_framework_input_converter.py | Removed Agent Framework adapter tests (package removal). |
| sdk/agentserver/azure-ai-agentserver-agentframework/tests/conftest.py | Removed Agent Framework adapter test harness. |
| sdk/agentserver/azure-ai-agentserver-agentframework/tests/init.py | Removed Agent Framework tests package marker. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/workflow_agent_simple/requirements.txt | Removed Agent Framework samples dependencies. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/workflow_agent_simple/.envtemplate | Removed Agent Framework sample env template. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/simple_async/requirements.txt | Removed Agent Framework sample deps. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/simple_async/minimal_async_example.py | Removed Agent Framework sample script. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/simple_async/README.md | Removed Agent Framework sample docs. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/simple_async/.envtemplate | Removed Agent Framework env template. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_simple/requirements.txt | Removed Agent Framework MCP sample deps. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_simple/mcp_simple.py | Removed Agent Framework MCP sample script. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_simple/.envtemplate | Removed Agent Framework MCP env template. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_apikey/requirements.txt | Removed Agent Framework MCP API-key sample deps. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_apikey/mcp_apikey.py | Removed Agent Framework MCP API-key sample script. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_apikey/README.md | Removed Agent Framework MCP API-key docs. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/mcp_apikey/.envtemplate | Removed Agent Framework MCP API-key env template. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/basic_simple/requirements.txt | Removed Agent Framework basic sample deps. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/basic_simple/minimal_example.py | Removed Agent Framework basic sample script. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/basic_simple/README.md | Removed Agent Framework basic sample docs. |
| sdk/agentserver/azure-ai-agentserver-agentframework/samples/basic_simple/.envtemplate | Removed Agent Framework basic env template. |
| sdk/agentserver/azure-ai-agentserver-agentframework/pyrightconfig.json | Removed Agent Framework pyright config. |
| sdk/agentserver/azure-ai-agentserver-agentframework/pyproject.toml | Removed Agent Framework packaging metadata. |
| sdk/agentserver/azure-ai-agentserver-agentframework/dev_requirements.txt | Removed Agent Framework dev requirements. |
| sdk/agentserver/azure-ai-agentserver-agentframework/cspell.json | Removed Agent Framework spell config. |
| sdk/agentserver/azure-ai-agentserver-agentframework/azure/ai/agentserver/agentframework/models/constants.py | Removed Agent Framework adapter runtime constants. |
| sdk/agentserver/azure-ai-agentserver-agentframework/azure/ai/agentserver/agentframework/models/agent_id_generator.py | Removed Agent Framework adapter helpers. |
| sdk/agentserver/azure-ai-agentserver-agentframework/azure/ai/agentserver/agentframework/models/agent_framework_input_converters.py | Removed Agent Framework input converter impl. |
| sdk/agentserver/azure-ai-agentserver-agentframework/azure/ai/agentserver/agentframework/agent_framework.py | Removed Agent Framework adapter runtime implementation. |
| sdk/agentserver/azure-ai-agentserver-agentframework/azure/ai/agentserver/agentframework/init.py | Removed Agent Framework package init/export surface. |
| sdk/agentserver/azure-ai-agentserver-agentframework/README.md | Removed Agent Framework package readme. |
| sdk/agentserver/azure-ai-agentserver-agentframework/CHANGELOG.md | Removed Agent Framework package changelog. |
Comment on lines
+52
to
+58
| self._provider: "ResponseProviderProtocol | None" = provider | ||
| self._input_items: list[InputParam] = list(input_items) if input_items is not None else [] | ||
| self._previous_response_id: str | None = previous_response_id | ||
| self.conversation_id: str | None = conversation_id | ||
| self._history_limit: int = history_limit | ||
| self._input_items_cache: Sequence[OutputItem] | None = None | ||
| self._history_cache: Sequence[OutputItem] | None = None |
There was a problem hiding this comment.
_input_items_cache is typed as Sequence[OutputItem], but get_input_items_async() caches and returns InputParam values. This is an internal type mismatch that will confuse type-checkers and can mask real bugs. Change _input_items_cache to Sequence[InputParam] | None.
Comment on lines
+29
to
+34
| def __init__(self) -> None: | ||
| """Initialise the subject with an empty event buffer and no subscribers.""" | ||
| self._events: list[dict[str, Any]] = [] | ||
| self._subscribers: list[asyncio.Queue[Any]] = [] | ||
| self._done: bool = False | ||
| self._lock: asyncio.Lock = asyncio.Lock() |
There was a problem hiding this comment.
_events (replay buffer) and each subscriber queue are unbounded, so a slow/paused SSE client can cause unbounded memory growth (potential OOM). Consider: (1) bounding the replay buffer (e.g., max events / max bytes), (2) using asyncio.Queue(maxsize=...) for subscribers, and (3) defining an overflow policy (drop subscriber, drop oldest, or apply backpressure) that matches SSE replay semantics.
Comment on lines
+36
to
+45
| async def publish(self, event: dict[str, Any]) -> None: | ||
| """Push a new event to all current subscribers and append it to the replay buffer. | ||
|
|
||
| :param event: The normalised event dict (``{"type": str, "payload": dict}``). | ||
| :type event: dict[str, Any] | ||
| """ | ||
| async with self._lock: | ||
| self._events.append(event) | ||
| for q in self._subscribers: | ||
| q.put_nowait(event) |
There was a problem hiding this comment.
Publishing while holding _lock and pushing directly into subscriber queues makes it difficult to introduce backpressure later (e.g., switching to await q.put(...) would block under the lock). A safer pattern is: take a snapshot (copy) of the subscriber list + append to _events under the lock, then perform queue writes outside the lock and handle overflow per-subscriber.
Comment on lines
+37
to
+52
| """Build a ``JSONResponse`` with the standard error envelope. | ||
|
|
||
| :param code: Machine-readable error code (e.g. ``"internal_error"``). | ||
| :type code: str | ||
| :param message: Human-readable error message. | ||
| :type message: str | ||
| :keyword status_code: HTTP status code for the response. | ||
| :paramtype status_code: int | ||
| :keyword error_type: Optional error type classification string. When | ||
| provided, included as ``"type"`` in the error body. | ||
| :paramtype error_type: Optional[str] | ||
| :keyword details: Child error objects, each with at least ``code`` and | ||
| ``message`` keys. | ||
| :paramtype details: Optional[list[dict[str, Any]]] | ||
| :keyword headers: Extra HTTP headers to include on the response. | ||
| :paramtype headers: Optional[dict[str, str]] |
There was a problem hiding this comment.
The docstring uses :paramtype which is not a standard Sphinx field and likely won’t render as intended. Replace with :type status_code: int (and similarly for the keyword-only fields) or use a consistent :keyword ...: / :type ...: pattern throughout.
| "Programming Language :: Python :: 3.11", | ||
| "Programming Language :: Python :: 3.12", | ||
| "Programming Language :: Python :: 3.13", | ||
| "Programming Language :: Python :: 3.14", |
There was a problem hiding this comment.
Programming Language :: Python :: 3.14 may not be a valid Trove classifier yet (depending on PyPI’s current classifier set). Invalid classifiers can cause publish/metadata validation failures. Consider removing 3.14 until it’s officially available in Trove classifiers, or gating it behind an explicit decision to support that runtime.
Suggested change
| "Programming Language :: Python :: 3.14", |
| "Programming Language :: Python :: 3.11", | ||
| "Programming Language :: Python :: 3.12", | ||
| "Programming Language :: Python :: 3.13", | ||
| "Programming Language :: Python :: 3.14", |
There was a problem hiding this comment.
Same concern as core: Programming Language :: Python :: 3.14 may not be a valid Trove classifier and can break packaging metadata validation during publish. Consider removing until officially supported.
Suggested change
| "Programming Language :: Python :: 3.14", |
Comment on lines
+3
to
+7
| ## 2.0.0b1 (Unreleased) | ||
|
|
||
| ### Features Added | ||
|
|
||
| First version | ||
| - `AgentHost` host framework with health probe (`/readiness`), graceful shutdown, and port binding. |
There was a problem hiding this comment.
The PR description/title state this is an independent review of the azure-ai-agentserver-responses package, but the diff also includes broad refactors in azure-ai-agentserver-core, introduces azure-ai-agentserver-invocations, and removes large portions of langgraph/agentframework packages. If this scope is intended, please update the PR title/description to match; otherwise, consider splitting these changes so the Responses package review isn’t blocked by unrelated removals/refactors.
Comment on lines
+11
to
+12
| class _ResponseEventSubject: | ||
| """In-process hot observable with replay buffer for SSE event broadcasting. |
There was a problem hiding this comment.
The Responses hosting changes introduce new concurrency/memory-sensitive behavior (replay buffer + subscriber management) but this PR doesn’t show corresponding unit tests for the new responses package. Add tests that cover: replay-from-cursor behavior, concurrent subscribe/unsubscribe during publish()/complete(), and completion behavior for late subscribers.
Comment on lines
+8
to
+72
| from typing import Any | ||
|
|
||
| from azure.ai.agentserver.responses.models._generated.sdk.models._types import InputParam | ||
|
|
||
| from .._response_context import ResponseContext | ||
|
|
||
|
|
||
| class _ExecutionContext: # pylint: disable=too-many-instance-attributes | ||
| """Holds all per-request *input* state for a single create-response call. | ||
|
|
||
| Passed between the routing layer and the orchestrator to avoid | ||
| large keyword-argument bundles on every internal function call. | ||
| All fields are set once at construction and treated as immutable by | ||
| the orchestrator. Mutable in-flight state (accumulated events, | ||
| background record, captured error) lives in :class:`_PipelineState` | ||
| inside the orchestrator methods. | ||
| """ | ||
|
|
||
| def __init__( | ||
| self, | ||
| *, | ||
| response_id: str, | ||
| agent_reference: Any, | ||
| model: str | None, | ||
| store: bool, | ||
| background: bool, | ||
| stream: bool, | ||
| input_items: list[InputParam], | ||
| previous_response_id: str | None, | ||
| conversation_id: str | None, | ||
| cancellation_signal: asyncio.Event, | ||
| span: Any, | ||
| parsed: Any, | ||
| agent_session_id: str | None = None, | ||
| context: ResponseContext | None = None, | ||
| ) -> None: | ||
| self.response_id = response_id | ||
| """The assigned response identifier.""" | ||
| self.agent_reference = agent_reference | ||
| """Normalized agent reference dictionary.""" | ||
| self.model = model | ||
| """Model name, or ``None`` if not specified.""" | ||
| self.store = store | ||
| """Whether to persist the execution record after the response completes.""" | ||
| self.background = background | ||
| """Whether this is a background (non-blocking) request.""" | ||
| self.stream = stream | ||
| """Whether to produce SSE streaming output.""" | ||
| self.input_items = input_items | ||
| """Extracted input items from the request body.""" | ||
| self.previous_response_id = previous_response_id | ||
| """Previous response ID for conversation chaining, or ``None``.""" | ||
| self.conversation_id = conversation_id | ||
| """Conversation ID for grouping related responses, or ``None``.""" | ||
| self.cancellation_signal = cancellation_signal | ||
| """Event signalling that the client has requested cancellation.""" | ||
| self.agent_session_id = agent_session_id | ||
| """Resolved session ID for this request (S-048).""" | ||
| self.context: ResponseContext | None = context | ||
| """Runtime response context for this request. Set after construction | ||
| via :meth:`_ResponseEndpointHandler._create_response_context`.""" | ||
| self.span = span | ||
| """Active observability span for this request.""" | ||
| self.parsed = parsed | ||
| """Parsed ``CreateResponse`` model instance from the request body.""" |
There was a problem hiding this comment.
_ExecutionContext is functioning as an immutable request-scoped record, but it’s implemented as a mutable class with many Any fields. Making it a @dataclass(frozen=True, slots=True) (or at least tightening types like agent_reference: dict[str, Any] and span to a concrete protocol/type) would reduce accidental mutation and improve readability/IDE support.
Suggested change
| from typing import Any | |
| from azure.ai.agentserver.responses.models._generated.sdk.models._types import InputParam | |
| from .._response_context import ResponseContext | |
| class _ExecutionContext: # pylint: disable=too-many-instance-attributes | |
| """Holds all per-request *input* state for a single create-response call. | |
| Passed between the routing layer and the orchestrator to avoid | |
| large keyword-argument bundles on every internal function call. | |
| All fields are set once at construction and treated as immutable by | |
| the orchestrator. Mutable in-flight state (accumulated events, | |
| background record, captured error) lives in :class:`_PipelineState` | |
| inside the orchestrator methods. | |
| """ | |
| def __init__( | |
| self, | |
| *, | |
| response_id: str, | |
| agent_reference: Any, | |
| model: str | None, | |
| store: bool, | |
| background: bool, | |
| stream: bool, | |
| input_items: list[InputParam], | |
| previous_response_id: str | None, | |
| conversation_id: str | None, | |
| cancellation_signal: asyncio.Event, | |
| span: Any, | |
| parsed: Any, | |
| agent_session_id: str | None = None, | |
| context: ResponseContext | None = None, | |
| ) -> None: | |
| self.response_id = response_id | |
| """The assigned response identifier.""" | |
| self.agent_reference = agent_reference | |
| """Normalized agent reference dictionary.""" | |
| self.model = model | |
| """Model name, or ``None`` if not specified.""" | |
| self.store = store | |
| """Whether to persist the execution record after the response completes.""" | |
| self.background = background | |
| """Whether this is a background (non-blocking) request.""" | |
| self.stream = stream | |
| """Whether to produce SSE streaming output.""" | |
| self.input_items = input_items | |
| """Extracted input items from the request body.""" | |
| self.previous_response_id = previous_response_id | |
| """Previous response ID for conversation chaining, or ``None``.""" | |
| self.conversation_id = conversation_id | |
| """Conversation ID for grouping related responses, or ``None``.""" | |
| self.cancellation_signal = cancellation_signal | |
| """Event signalling that the client has requested cancellation.""" | |
| self.agent_session_id = agent_session_id | |
| """Resolved session ID for this request (S-048).""" | |
| self.context: ResponseContext | None = context | |
| """Runtime response context for this request. Set after construction | |
| via :meth:`_ResponseEndpointHandler._create_response_context`.""" | |
| self.span = span | |
| """Active observability span for this request.""" | |
| self.parsed = parsed | |
| """Parsed ``CreateResponse`` model instance from the request body.""" | |
| from dataclasses import dataclass | |
| from typing import Any | |
| from azure.ai.agentserver.responses.models._generated.sdk.models._types import InputParam | |
| from .._response_context import ResponseContext | |
| @dataclass(frozen=True, slots=True) # pylint: disable=too-many-instance-attributes | |
| class _ExecutionContext: | |
| """Holds all per-request *input* state for a single create-response call. | |
| Passed between the routing layer and the orchestrator to avoid | |
| large keyword-argument bundles on every internal function call. | |
| All fields are set once at construction and enforced as immutable by | |
| the dataclass configuration. Mutable in-flight state (accumulated events, | |
| background record, captured error) lives in :class:`_PipelineState` | |
| inside the orchestrator methods. | |
| """ | |
| response_id: str | |
| """The assigned response identifier.""" | |
| agent_reference: dict[str, Any] | |
| """Normalized agent reference dictionary.""" | |
| model: str | None | |
| """Model name, or ``None`` if not specified.""" | |
| store: bool | |
| """Whether to persist the execution record after the response completes.""" | |
| background: bool | |
| """Whether this is a background (non-blocking) request.""" | |
| stream: bool | |
| """Whether to produce SSE streaming output.""" | |
| input_items: list[InputParam] | |
| """Extracted input items from the request body.""" | |
| previous_response_id: str | None | |
| """Previous response ID for conversation chaining, or ``None``.""" | |
| conversation_id: str | None | |
| """Conversation ID for grouping related responses, or ``None``.""" | |
| cancellation_signal: asyncio.Event | |
| """Event signalling that the client has requested cancellation.""" | |
| span: Any | |
| """Active observability span for this request.""" | |
| parsed: Any | |
| """Parsed ``CreateResponse`` model instance from the request body.""" | |
| agent_session_id: str | None = None | |
| """Resolved session ID for this request (S-048).""" | |
| context: ResponseContext | None = None | |
| """Runtime response context for this request. Set after construction | |
| via :meth:`_ResponseEndpointHandler._create_response_context`.""" | |
| def _test_execution_context_frozen_semantics() -> None: | |
| """Basic test to validate that :class:`_ExecutionContext` is immutable. | |
| This function constructs an instance of :class:`_ExecutionContext` and | |
| verifies that attempting to mutate an attribute raises | |
| :class:`dataclasses.FrozenInstanceError`. It is intended to provide | |
| lightweight coverage for the frozen dataclass behavior. | |
| """ | |
| from dataclasses import FrozenInstanceError | |
| ctx = _ExecutionContext( | |
| response_id="response-id", | |
| agent_reference={}, | |
| model=None, | |
| store=True, | |
| background=False, | |
| stream=False, | |
| input_items=[], | |
| previous_response_id=None, | |
| conversation_id=None, | |
| cancellation_signal=asyncio.Event(), | |
| span=object(), | |
| parsed=object(), | |
| agent_session_id=None, | |
| context=None, | |
| ) | |
| try: | |
| # Attempt to mutate a frozen dataclass field should raise. | |
| ctx.response_id = "other-id" # type: ignore[misc] | |
| except FrozenInstanceError: | |
| # Expected path: immutability is enforced. | |
| return | |
| raise AssertionError( | |
| "_ExecutionContext must be immutable, but attribute assignment succeeded." | |
| ) |
Comment on lines
+26
to
+30
| def __init__( | ||
| self, | ||
| *, | ||
| response_id: str, | ||
| agent_reference: Any, |
There was a problem hiding this comment.
_ExecutionContext is functioning as an immutable request-scoped record, but it’s implemented as a mutable class with many Any fields. Making it a @dataclass(frozen=True, slots=True) (or at least tightening types like agent_reference: dict[str, Any] and span to a concrete protocol/type) would reduce accidental mutation and improve readability/IDE support.
API Change CheckAPIView identified API level changes in this PR and created the following API reviews |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Split from PR #45925 for independent review of the
azure-ai-agentserver-responsespackage.This PR builds on the core+invocations changes in #45925 and adds the responses protocol implementation including:
azure.core.AsyncPipelineClient, in-memory provider for testingDependencies
agentserver/invoke-reponses(PR [agentserver] azure-ai-agentserver -core and -invocation packages #45925) which provides the core packageCarried-over review threads from #45925
The following unresolved review threads were moved from the original PR. They all target code in this package.
Code fixes needed (copilot-reviewer)
_http_errors.py_not_foundreturns 404 but usescode="invalid_request"— should usecode="not_found"anderror_type="not_found_error"_handlers.pyresponse_handlervalidates parameter count but does not validate handler isasync defor returnsAsyncIterable_foundry_serializer.pyserialize_create_requestcalls.as_dict()unconditionally — breaks if adictis passed instead of a model_base.pyResponseProviderProtocoldocstrings specifyKeyErrorbut Foundry provider raisesFoundryResourceNotFoundError(2 threads, duplicate)_runtime_state.py[*A, *history]— use deque or accumulate-then-concat_event_subject.pyasyncio.Queueper subscriber with no backpressure — can cause OOM (2 threads, duplicate)_foundry_settings.pybuild_url()uses manual string concatenation — switch tourllib.parse.urlencode_request_parsing.pyDesign/architecture (johanste)
_routing.pyResponseHandlershould inherit fromAgentHostto be the host entry pointsamples/GetStarted/app.pysamples/GetStarted/app.pyResponseHandlershould be the "host" — users should not needAgentHostdirectlySamples (RaviPidaparthi)
samples/README.md