Compare commits

..
Author SHA1 Message Date
Evan MattsonandGitHub 26cd5cc1bf Python: Bump Python version to 1.0.0rc5 and 1.0.0b260319 for a release. (#4807)
* Bump Python version to 1.0.0rc5 and 1.0.0b260319 for a release.

* update missed pkg versions

* Update changelog
2026-03-20 10:35:47 +09:00
James SturtevantandGitHub b4c4f5094e Python: Emit tool call events in GitHubCopilotAgent streaming (#4711)
* Emit tool call events in GitHubCopilotAgent streaming

_stream_updates now yields FunctionCallContent for TOOL_EXECUTION_START
and FunctionResultContent for TOOL_EXECUTION_COMPLETE events from the
Copilot SDK session. This enables DevUI and other consumers to display
tool calls during streaming agent execution. Previously only ASSISTANT_MESSAGE_DELTA, SESSION_IDLE, and SESSION_ERROR
were handled — tool execution events were silently dropped.

Signed-off-by: James Sturtevant <jsturtevant@gmail.com>

* Add some tests

Signed-off-by: James Sturtevant <jsturtevant@gmail.com>

* Respond to feedback

Signed-off-by: James Sturtevant <jsturtevant@gmail.com>

* Fix TOOL_EXECUTION_COMPLETE to use correct SDK types

- Read result text from session_events.Result.content (not ToolResult.text_result_for_llm)
- Read failure state from event.data.success/error (not result_obj.result_type/error)
- Handle ErrorClass.message and plain string errors
- Update tests to use session_events.Result and ErrorClass
- Add tests for string errors, success-with-error, and COMPLETE missing fields

Signed-off-by: James Sturtevant <jsturtevant@gmail.com>

---------

Signed-off-by: James Sturtevant <jsturtevant@gmail.com>
2026-03-20 01:27:36 +00:00
0cd40f8354 Python: [BREAKING] Refactor middleware layering and split Anthropic raw client (#4746)
* [BREAKING] Refactor middleware layering and raw clients

Reorder chat client layers so function invocation wraps chat middleware, and chat middleware stays outside telemetry while still running for each inner model call. Add middleware pipeline caching, refresh docs and samples, and split Anthropic into raw and public clients to match the standard layering model.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Tighten typing ignores in ancillary modules

Add targeted typing ignores in workflow visualization and lab modules so pyright stays clean alongside the middleware refactor work.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix categorize_middleware to unpack tuple/Sequence and use relative MRO assertions

- Broaden isinstance check in categorize_middleware from list to Sequence
  so tuples and other Sequence types are properly unpacked instead of
  being appended as a single item.
- Replace fragile hardcoded MRO index assertions in anthropic test with
  relative ordering via mro.index().
- Add regression tests for categorize_middleware with tuple, list, and
  None inputs.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix middleware string decomposition, add middleware param to FunctionInvocationLayer, and add tests (#4710)

- Guard categorize_middleware Sequence check against str/bytes to prevent
  character-by-character decomposition of accidentally passed strings
- Add explicit middleware parameter to FunctionInvocationLayer.get_response
  and merge it into client_kwargs before categorization, fixing the
  inconsistency where only OpenAIChatClient supported this parameter
- Add assertions that RawAnthropicClient does not inherit convenience layers
- Add chat middleware cache test with non-empty base middleware
- Add tests for single unwrapped middleware item and string input

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Apply pre-commit auto-fixes

* Address review feedback for #4710: review comment fixes

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <copilot@github.com>
2026-03-20 00:43:37 +00:00
cefda44283 Python: Emit TOOL_CALL_RESULT events when resuming after tool approval (#4758)
* Emit TOOL_CALL_RESULT events on approval resume (#4589)

When a tool call is approved via the interrupt/resume flow,
_resolve_approval_responses executes the tool and injects the result
into the messages array, but no TOOL_CALL_RESULT SSE event was yielded
to the client.

Changes:
- _resolve_approval_responses now returns the list of resolved
  function_result Content objects instead of None
- run_agent_stream yields ToolCallResultEvent for each resolved
  approval result after RunStartedEvent is emitted
- Add ToolCallResultEvent to ag_ui.core imports in _agent_run.py

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* fix(ag-ui): address PR review feedback for #4589

1. _resolve_approval_responses now returns only approved results (not
   rejections) so TOOL_CALL_RESULT events are emitted only for executed
   tools. Rejection results are still written into message history.

2. Emit resolved TOOL_CALL_RESULT events in the no-updates fallback
   RUN_STARTED path so approval results are never lost.

3. Rewrite tests to use real FunctionTool with func and
   approval_mode='always_require' via StubAgent default_options,
   verifying actual tool execution output in TOOL_CALL_RESULT content.
   Added test for rejection not emitting TOOL_CALL_RESULT.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix #4589: clean up approval resolution and add missing tests

- Extract duplicated TOOL_CALL_RESULT emission block into
  _make_approval_tool_result_events helper to prevent drift
- Remove dead rejection_results construction in _resolve_approval_responses;
  _replace_approval_contents_with_results already handles rejections inline
- Pass only approved_results (not all_results) to clarify the contract
- Add mixed approve/reject test validating the core splitting logic
- Add zero-updates test covering the no-updates fallback emission path
- Add direct unit test for _resolve_approval_responses return value

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Fix import sorting lint error in test_approval_result_event.py

Add blank line between first-party and third-party import groups
to satisfy ruff I001 rule.

Fixes #4589

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-20 00:41:46 +00:00
4afc088f01 Python: Emit AG-UI events for MCP tool calls, results, and text reasoning (#4760)
* Python: Emit AG-UI events for MCP tool calls, results, and text reasoning

Fixes #4213 — `_emit_content()` in the AG-UI layer only handled `text`,
`function_call`, `function_result`, `function_approval_request`, `usage`,
and `oauth_consent_request` content types. Foundry MCP content types
(`mcp_server_tool_call`, `mcp_server_tool_result`) and `text_reasoning`
fell through unhandled, producing no SSE events for AG-UI consumers.

Added three new handler functions wired into `_emit_content()`:

- `_emit_mcp_tool_call`: emits TOOL_CALL_START + TOOL_CALL_ARGS and
  tracks in FlowState for MESSAGES_SNAPSHOT inclusion
- `_emit_mcp_tool_result`: emits TOOL_CALL_END + TOOL_CALL_RESULT with
  full FlowState cleanup mirroring `_emit_tool_result`
- `_emit_text_reasoning`: emits the protocol-defined reasoning event
  sequence (ReasoningStart → MessageStart → MessageContent → MessageEnd
  → ReasoningEnd) with ReasoningEncryptedValueEvent for protected_data

* Add HTTP round-trip tests for MCP tool and reasoning SSE events

Exercises the full POST → SSE bytes → parse → validate pipeline for
mcp_server_tool_call, mcp_server_tool_result, text_reasoning, and
ReasoningEncryptedValueEvent content through FastAPI TestClient.

* Fix _emit_mcp_tool_result missing predictive_handler support (#4213)

- Add predictive_handler parameter to _emit_mcp_tool_result and mirror
  the apply_pending_updates + StateSnapshotEvent block from _emit_tool_result
- Forward predictive_handler from _emit_content to _emit_mcp_tool_result
- Add assertion for stored arguments in MCP tool call test
- Add test for predictive handler state snapshot after MCP tool result

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Refactor MCP tool emit functions and add missing tests (#4213)

- Extract _emit_tool_result_common shared helper to eliminate duplication
  between _emit_tool_result and _emit_mcp_tool_result
- Remove server_name prefix from tool_call_name in _emit_mcp_tool_call;
  display_name now equals tool_name directly
- Add test for tool_name fallback to 'mcp_tool' when tool_name is None
- Add test for output=None fallback to empty string in _emit_mcp_tool_result

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address review feedback for #4213: review comment fixes

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-20 00:41:37 +00:00
1272ec5adf Add automated stale issue and PR follow-up ping workflow (#4776)
* Add script to ping on stale issues/PRs

* Add script to ping on stale issues/PRs

* Fix stale issue/PR ping script review comments

- Rename TEAM_NAME env var to TEAM_SLUG for clarity
- Add actionable error messages for 403/404 team lookup failures
- Add contents:read permission for actions/checkout
- Use github.event.inputs context with fallback for scheduled runs
- Pin PyGithub to 2.6.0 for reproducible builds
- Fetch comments once in should_ping() to reduce API calls
- Make ping() retry loop idempotent (track comment/label state)
- Validate DAYS_THRESHOLD with helpful error for non-numeric input
- Fix timezone bug: use astimezone() instead of replace(tzinfo=)
- Add comprehensive unit tests (29 tests)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-20 00:41:31 +00:00
47ead84753 Python: Support detail field in OpenAI Chat API image_url payload (#4756)
* Support detail field in OpenAI image_url payload (#4616)

Include the optional 'detail' field from Content.additional_properties
when building image_url payloads for the OpenAI Chat API, matching the
existing pattern used for 'filename' in document file payloads.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Remove reproduction report

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Simplify detail extraction from additional_properties (#4616)

- Remove unnecessary hasattr check; additional_properties is always
  initialized as a dict on Content instances.
- Use 'is not None' instead of truthy check to be more precise.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Remove detail allowlist in chat client to align with responses client

Replace the strict allowlist check ('low', 'high', 'auto') with an
isinstance(detail, str) check so that any valid string detail value is
passed through to OpenAI. This aligns the chat client behavior with the
responses client, which passes detail through unconditionally.

Also add test coverage for:
- Future/unknown string detail values being passed through
- Data URI images (covering the 'data' branch of the match)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-20 00:33:57 +00:00
4c287c2424 Python: Fix MCP tool schema normalization for zero-argument tools missing 'properties' key (#4771)
* Fix zero-argument MCP tool schema missing 'properties' key (#4540)

MCP servers for zero-argument tools (e.g. matlab-mcp-core-server's
detect_matlab_toolboxes) declare inputSchema as {"type": "object"}
without a "properties" key. OpenAI's API requires "properties" to
be present on object schemas, causing a 400 invalid_request_error.

Normalize inputSchema at MCP ingestion in load_tools() to inject an
empty "properties": {} when it is missing from object-type schemas.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address review feedback for #4540: improve test robustness and add defensive guard

- Look up loaded functions by name instead of index to avoid brittle
  ordering assumptions
- Add negative-path test cases: non-object schema (type: string) and
  empty schema ({}) to verify guard clause skips them correctly
- Assert original inputSchema dicts are not mutated by load_tools()
- Add defensive guard for tool.inputSchema being None

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address review feedback for #4540: Python: [Bug]: Local stdio MCP works for calculator but fails for official matlab-mcp-core-server on LM Studio /v1/responses

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-19 22:13:37 +00:00
westeyandGitHub 100086a276 Add docker-in-docker feature to dev container (#4794) 2026-03-19 19:18:46 +00:00
fc6721ca8e .NET: Trim src references and add utility to enforce (#4693)
* Trim src references and add utility to enforce

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
2026-03-19 10:57:43 +00:00
4b21f38650 Python: Fix A2AAgent to invoke context providers before and after run (#4757)
* Fix A2AAgent to invoke context providers before and after run

A2AAgent.run() bypassed the context provider lifecycle (before_run/after_run)
that BaseAgent defines as a contract for all agents. This caused A2AAgent to
violate the semantic definition of BaseAgent, resulting in inconsistency with
other agent implementations.

The fix follows the same pattern used by WorkflowAgent:
- Create SessionContext and run before_run on all context providers before
  processing the A2A stream
- Collect response updates and run after_run on all context providers after
  the stream is fully consumed
- Auto-create a session when context providers are configured but no session
  is explicitly passed

Fixes #4754

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Remove reproduction report from repository

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address PR review feedback for #4754

- Validate messages when no continuation_token: raise ValueError if
  normalized_messages is empty, preventing IndexError on messages[-1]
- Import BaseContextProvider/SessionContext from public agent_framework
  package instead of internal agent_framework._sessions module
- Add test for ValueError on run(None) without continuation_token

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Improve test coverage for empty-messages guard in A2AAgent.run (#4754)

- Parameterize test to cover both messages=None and messages=[] inputs
- Add test verifying run(None, continuation_token=...) does not raise

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-19 10:45:42 +00:00
bf8d9672e1 Python: Aggregate token usage across tool-call loop iterations in invoke_agent span (#4739)
* Fix invoke_agent span to aggregate token usage across LLM calls (#4062)

The FunctionInvocationLayer._get_response() loop was overwriting the
response on each iteration, so usage_details only reflected the last
chat completion call. Now tracks aggregated_usage across all iterations
using add_usage_details() and sets it on the returned response.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Remove reproduction report artifact

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Apply pre-commit auto-fixes

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-19 06:41:33 +00:00
Peter IbekweandGitHub 5374dd47c5 .NET: Fix source generator bug that silently drops base class handler registrations for protocol-only partial executors (#4751)
* Fix source generator bug that silently drops base class handler registrations for protocol-only partial executors

* Fixed xml comments and variable naming.
2026-03-19 02:19:42 +00:00
29dfcbb584 .NET: Validate SkillsInstructionPrompt contains {0} placeholder in FileAgentSkillsProvider (#4642)
* Fix FileAgentSkillsProvider accepting SkillsInstructionPrompt without {0} placeholder (#4638)

BuildSkillsInstructionPrompt validated only format-string syntax via
string.Format(template, ""), which silently accepted templates without a
{0} placeholder. The generated skills list was then dropped from the final
instructions.

Tighten validation to format with a sentinel string and verify it appears
in the output, rejecting templates that do not reference argument 0 with
an ArgumentException.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix netstandard2.0 compat and simplify prompt template validation (#4638)

- Replace string.Contains(string, StringComparison) with IndexOf for
  netstandard2.0/net472 compatibility
- Remove sentinel round-trip check; validate {0} directly on the raw
  template string using IndexOf
- Add positive test verifying custom SkillsInstructionPrompt with {0}
  is accepted and applied to output

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-19 00:18:39 +00:00
CopilotGitHubcrickmanCopilot Autofix powered by AIcopilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>Chris Rickman
c9321b9028 .NET Compaction - Allow developer to specify a custom formatter for ToolResultCompactionStrategy (#4667)
* Initial plan

* Allow developer to specify custom formatter for ToolResultCompactionStrategy

Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>

* Refine shape

* Fix test expectation

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>
Co-authored-by: Chris Rickman <crickman@microsoft.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
2026-03-18 19:24:43 +00:00
f48c4512d3 Python: Simplify Python Poe tasks and unify package selectors (#4722)
* updated automation tasks and commands, with alias for the time being

* Restore aggregate test exclusions

Preserve the legacy all-tests scope for test --all by excluding lab and devui from the default aggregate sweep, while still allowing explicit package selection. Also ignore hidden/generated test directories such as .mypy_cache during aggregate discovery.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* updated versions in pre-commit

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-18 18:39:11 +00:00
CopilotGitHubcrickmanCopilot Autofix powered by AIcopilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>Chris Rickman
d3d0100822 .NET Compaction - Add AsChatReducer() extension to expose CompactionStrategy as IChatReducer (#4664)
* Initial plan

* Add ChatStrategyExtensions.cs with AsChatReducer() extension method and tests

Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>

* Refactor message list creation in ReduceAsync method

* Remove unnecessary blank line in AsChatReducer method

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

* Fix test

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Chris Rickman <crickman@microsoft.com>
2026-03-18 17:25:54 +00:00
acaf6b7054 Python: Fix ENABLE_SENSITIVE_DATA env var ignored when set after module import (#4743)
* Python: Re-read env vars in configure_otel_providers and enable_instrumentation (#4119)

Fix ENABLE_SENSITIVE_DATA and VS_CODE_EXTENSION_PORT env vars being ignored
when load_dotenv() runs after module import. The module-level
OBSERVABILITY_SETTINGS singleton cached env state at import time, and
configure_otel_providers() / enable_instrumentation() never re-read from
os.environ when parameters were None.

Both functions now construct a fresh ObservabilitySettings() to pick up
current env vars when explicit parameters are not provided, matching the
existing behavior of the env_file_path branch.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address PR review feedback for #4119: avoid throwaway ObservabilitySettings

- Add _read_bool_env/_read_int_env helpers to read env vars without
  constructing a full ObservabilitySettings (which calls create_resource())
- Replace ObservabilitySettings() in enable_instrumentation() and
  configure_otel_providers() else-branch with direct env reads
- Add enable_console_exporters parameter to configure_otel_providers()
  for override parity with enable_sensitive_data and vs_code_extension_port
- Propagate _resource and _executed_setup in the non-env_file_path branch
- Make existing tests hermetic (clear VS_CODE_EXTENSION_PORT and
  ENABLE_CONSOLE_EXPORTERS env vars)
- Add tests: enable_console_exporters env refresh, explicit param overrides
  for both enable_instrumentation() and configure_otel_providers()

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address remaining review feedback for #4119

- Refresh enable_console_exporters in enable_instrumentation() for
  consistency with configure_otel_providers(), so env var changes
  after import are picked up by both public API functions
- Make test_configure_otel_providers_reads_env_vs_code_port hermetic
  by clearing ENABLE_CONSOLE_EXPORTERS from the environment
- Add test_enable_instrumentation_reads_env_console_exporters to
  cover the new refresh behavior

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Remove unconditional enable_console_exporters overwrite from enable_instrumentation() (#4119)

enable_instrumentation() is documented as not configuring exporters, so
managing enable_console_exporters there was a leaky abstraction. The
unconditional _read_bool_env call silently reset the value to False when
ENABLE_CONSOLE_EXPORTERS was absent from env, clobbering any value
previously set by configure_otel_providers(enable_console_exporters=True).

- Remove the unconditional overwrite line from enable_instrumentation()
- Replace test_enable_instrumentation_reads_env_console_exporters with
  test_enable_instrumentation_does_not_touch_console_exporters
- Add regression test: enable_instrumentation() does not clobber a
  previously configured enable_console_exporters value
- Add test: explicit enable_sensitive_data param still leaves
  enable_console_exporters untouched

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-18 15:58:22 +00:00
Hui MiaoandGitHub c2fec6b51c Python: Add foundry hosted agents samples for python (#4648)
* Add two hosted agent samples using the foundry agent

* Refactor formatting and improve readability in main.py

* Add agent-framework dependency to requirements and update copyright notice in main.py files

* Refactor agent imports and update credential handling in hosted agent samples

* Update agent framework dependency in requirements for hosted agents

* chore: update Python version to 3.14 and improve Dockerfile for hosted agents

* feat: add hosted agent samples for Azure AI with local tools and multi-agent workflows

* fix: update Azure AI client import and refactor agent initialization in hotel agent sample

* feat: add hosted agent samples for Seattle hotel search and writer-reviewer workflow

* fix: correct agent name in YAML configuration for local tools agent
2026-03-18 08:39:08 +00:00
705ed47a0b Python: Fix missing methods on the Content class in durable tasks (#4738)
* Fix Content serialization in DurableAgentStateUnknownContent (#4719)

DurableAgentStateUnknownContent.from_unknown_content() stored raw Content
objects without converting them to dicts, causing json.dumps to fail in
Azure Durable Functions' entity state serialization. This affected content
types not explicitly handled (e.g., mcp_server_tool_call/result).

The fix converts Content objects to dicts via to_dict() when storing in
DurableAgentStateUnknownContent, and restores them via Content.from_dict()
in to_ai_content().

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Add to_json and from_json methods to Content class (#4719)

Add to_json() and from_json() methods to the Content class to match the
serialization interface provided by SerializationMixin on other model classes.
Also fix pre-existing pyright type errors in durabletask's
DurableAgentStateUnknownContent.to_ai_content().

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address PR review: add type guard, remove to_json, add fallback, and tests

- Remove Content.to_json() per reviewer request (comment 3)
- Add type guard in Content.from_json() for non-dict JSON (comments 1, 4)
- Wrap json.JSONDecodeError as ValueError for consistent exception contract
- Add try/except fallback in to_ai_content() for invalid Content dicts (comment 5)
- Add test_content_to_dict_exclude_none and test_content_to_dict_exclude_fields (comment 2)
- Add test_unknown_content_to_ai_content_fallback_on_invalid_type_dict (comment 5)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Address review feedback for #4719: review comment fixes

* Remove Content.from_json, move logic to consuming code (#4719)

Remove the from_json convenience method from Content class per review
feedback. This is the same trivial json.loads + from_dict wrapper as
to_json which was already removed. Consumers should call json.loads
and Content.from_dict directly.

Update tests to use Content.from_dict(json.loads(...)) pattern and
remove from_json-specific error handling tests (those errors are
already covered by json.loads and Content.from_dict).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-18 08:08:44 +00:00
192a283c9a Python: Reduce Azure chat client import overhead (#4744)
* Reduce Azure chat client import overhead

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix Azure chat client type annotations and add _parse_text_from_openai tests

- Move Choice and ChunkChoice imports under TYPE_CHECKING to avoid
  runtime import cost (from __future__ annotations is already present)
- Restore proper typed signature (Choice | ChunkChoice) instead of Any
- Add direct unit tests for _parse_text_from_openai covering:
  - Choice with message content
  - ChunkChoice with delta content
  - Refusal branch for both Choice and ChunkChoice
  - No content/no refusal returning None
  - None delta (async content filtering) returning None

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <copilot@github.com>
2026-03-18 08:05:42 +00:00
Peter IbekweandGitHub c74b1b08eb .NET: Fix race condition issue in FanInEdge while processing messages. (#4662)
* Fix race condition issue in FanInEdge while processing messages.

* refactored to limit the code segment under lock.

* Remove extra materialization of the result.

* Added comment to clarify future changes if process message is made async.
2026-03-18 00:36:10 +00:00
Shyju KrishnankuttyandGitHub 7c85f98c27 .NET: Align sample build configuration with test runner in CI (#4735)
* Run azure functions integration tests in release mode.

* Use debug when in debug build.
2026-03-17 20:20:14 +00:00
dependabot[bot]GitHubdependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
1e6f8909ec Bump pyjwt from 2.11.0 to 2.12.0 in /python (#4699)
Bumps [pyjwt](https://github.com/jpadilla/pyjwt) from 2.11.0 to 2.12.0.
- [Release notes](https://github.com/jpadilla/pyjwt/releases)
- [Changelog](https://github.com/jpadilla/pyjwt/blob/master/CHANGELOG.rst)
- [Commits](https://github.com/jpadilla/pyjwt/compare/2.11.0...2.12.0)

---
updated-dependencies:
- dependency-name: pyjwt
  dependency-version: 2.12.0
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-03-17 16:06:07 +00:00
dependabot[bot]GitHubdependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
008fe23585 Bump actions/upload-artifact from 4 to 7 (#4373)
Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4 to 7.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/v4...v7)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-version: '7'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-03-17 16:05:55 +00:00
dependabot[bot]GitHubdependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
6af0511e2b Bump MishaKav/pytest-coverage-comment from 1.2.0 to 1.6.0 (#4543)
Bumps [MishaKav/pytest-coverage-comment](https://github.com/mishakav/pytest-coverage-comment) from 1.2.0 to 1.6.0.
- [Release notes](https://github.com/mishakav/pytest-coverage-comment/releases)
- [Changelog](https://github.com/MishaKav/pytest-coverage-comment/blob/main/CHANGELOG.md)
- [Commits](https://github.com/mishakav/pytest-coverage-comment/compare/v1.2.0...v1.6.0)

---
updated-dependencies:
- dependency-name: MishaKav/pytest-coverage-comment
  dependency-version: 1.6.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-03-17 16:04:37 +00:00
dependabot[bot]GitHubdependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
6dbb0a5bb4 Bump danielpalme/ReportGenerator-GitHub-Action from 5.5.1 to 5.5.3 (#4542)
Bumps [danielpalme/ReportGenerator-GitHub-Action](https://github.com/danielpalme/reportgenerator-github-action) from 5.5.1 to 5.5.3.
- [Release notes](https://github.com/danielpalme/reportgenerator-github-action/releases)
- [Commits](https://github.com/danielpalme/reportgenerator-github-action/compare/5.5.1...5.5.3)

---
updated-dependencies:
- dependency-name: danielpalme/ReportGenerator-GitHub-Action
  dependency-version: 5.5.3
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-03-17 16:04:20 +00:00
dependabot[bot]GitHubdependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
21af304c7d Bump actions/setup-dotnet from 5.1.0 to 5.2.0 (#4541)
Bumps [actions/setup-dotnet](https://github.com/actions/setup-dotnet) from 5.1.0 to 5.2.0.
- [Release notes](https://github.com/actions/setup-dotnet/releases)
- [Commits](https://github.com/actions/setup-dotnet/compare/v5.1.0...v5.2.0)

---
updated-dependencies:
- dependency-name: actions/setup-dotnet
  dependency-version: 5.2.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-03-17 16:04:07 +00:00
94af83680e Python: Fix RUN_FINISHED.interrupt to accumulate all interrupts when multiple tools need approval (#4717)
* Fix flow.interrupts overwrite when multiple tools need approval (#4590)

Change flow.interrupts assignment to append so that all interrupt entries
accumulate when multiple tools require approval in a single turn.

Both _run_common.py and _agent_run.py used assignment (=) which caused
each new interrupt to overwrite the previous one. Switching to append()
ensures RUN_FINISHED.interrupt contains all pending approvals.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Add test for streaming path with multiple confirm_changes interrupts (#4590)

Add integration test exercising run_agent_stream with multiple predictive
tool calls requiring confirmation. Verifies that flow.interrupts.append()
correctly accumulates all interrupt entries and they appear in the
RUN_FINISHED event.

Also confirms FlowState already declares interrupts field with
default_factory=list, addressing the AttributeError concern from review.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-17 12:44:44 +00:00
cdb51e6a41 Python: fix thread serialization for multi-turn tool calls (#4684)
* Python: strip fc_id from loaded history

* Move fc_id replay handling into Responses client

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Remove unnecessary pytest asyncio marker

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Add Responses integration test for fc_id replay

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* removed old arg

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-17 10:00:04 +00:00
cbcdb2d29e .NET: Add durable workflow support (#4436)
* .NET: [Feature Branch] Add basic durable workflow support (#3648)

* Add basic durable workflow support.

* PR feedback fixes

* Add conditional edge sample.

* PR feedback fixes.

* Minor cleanup.

* Minor cleanup

* Minor formatting improvements.

* Improve comments/documentation on the execution flow.

* .NET: [Feature Branch] Add Azure Functions hosting support for durable workflows (#3935)

* Adding azure functions workflow support.

* - PR feedback fixes.
- Add example to demonstrate complex Object as payload.

* rename instanceId to runId.

* Use custom ITaskOrchestrator to run orchestrator function.

* .NET: [Feature Branch] Adding support for events & shared state in durable workflows (#4020)

* Adding support for events & shared state in durable workflows.

* PR feedback fixes

* PR feedback fixes.

* Add YieldOutputAsync calls to 05_WorkflowEvents sample executors

The integration test asserts that WorkflowOutputEvent is found in the
stream, but the sample executors only used AddEventAsync for custom
events and never called YieldOutputAsync. Since WorkflowOutputEvent is
only emitted via explicit YieldOutputAsync calls, the assertion would
fail. Added YieldOutputAsync to each executor to match the test
expectation and demonstrate the API in the sample.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix deserialization to use shared serializer options.

* PR feedback updates.

* Sample cleanup

* PR feedback fixes

* Addressing PR review feedback for DurableStreamingWorkflowRun

   - Use -1 instead of 0 for taskId in TaskFailedException when task ID is not relevant.
   - Add [NotNullWhen(true)] to TryParseWorkflowResult out parameter following .NET TryXXX conventions.

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* .NET: [Feature Branch]  Add nested sub-workflow support for durable workflows (#4190)

* .NET: [Feature Branch] Add nested sub-workflow support for durable workflows

* fix readme path

* Switch Orchestration output from string to DurableWorkflowResult.

* PR feedback fixes

* Minor cleanup based on PR feedback.

* .NET: [Feature Branch] Add Human In the Loop support for durable workflows (#4358)

* Add Azure Functions HITL workflow sample

Add 06_WorkflowHITL Azure Functions sample demonstrating Human-in-the-Loop
workflow support with HTTP endpoints for status checking and approval responses.

The sample includes:
- ExpenseReimbursement workflow with RequestPort for manager approval
- Custom HTTP endpoint to check workflow status and pending approvals
- Custom HTTP endpoint to send approval responses via RaiseEventAsync
- demo.http file with step-by-step interaction examples

* PR feedback fixes

* Minor comment cleanup

* Minor comment clReverted the `!context.IsReplaying` guards on `PendingEvents.Add`/`RemoveAll` and `SetCustomStatus` in `ExecuteRequestPortAsync`. The guards broke fan-out scenarios where parallel RequestPorts      need to be discoverable after replay. `SetCustomStatus` is idempotent metadata that doesn't affect replay determinism.eanup

* fix  for PR feedback

* PR feedback updates

* Improvements to samples

* Improvements to README

* Update samples to use parallel request ports.

* Unit tests

* Introduce local variables to improve readability of Workflows.Workflows access patter

* Use GitHub-style callouts and add PowerShell command variants in HITL sample README

* Add changelog entries for durable workflow support (#4436)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Bump Microsoft.DurableTask.Worker to 1.19.1 to fix version downgrade

Microsoft.Azure.Functions.Worker.Extensions.DurableTask 1.13.1 requires
Microsoft.DurableTask.Worker >= 1.19.1 via its transitive dependency on
Microsoft.DurableTask.Worker.Grpc 1.19.1.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix broken markdown links in durable workflow sample READMEs

- Create Workflow/README.md with environment setup docs
- Fix ../README.md -> ../../README.md in ConsoleApps 01, 02, 03, 08
- Fix SubWorkflows relative path (3 levels -> 4 levels up)
- Fix dead Durable Task Scheduler URL

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix build errors from main merge: Throw conflict, ExecuteAsync rename, GetNewSessionAsync rename

- Remove InjectSharedThrow from DurableTask csproj (uses Workflows' internal Throw via InternalsVisibleTo)
- Update ExecuteAsync -> ExecuteCoreAsync with WorkflowTelemetryContext.Disabled
- Update GetNewSessionAsync -> CreateSessionAsync

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Move durable workflow samples to 04-hosting/DurableWorkflows

Aligns with main branch sample reorganization where durable samples
live under 04-hosting/ (alongside DurableAgents/).

- Move samples/Durable/Workflow/ -> samples/04-hosting/DurableWorkflows/
- Add Directory.Build.props matching DurableAgents pattern
- Update slnx project paths
- Update integration test sample paths
- Update README cd paths and cross-references

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix build errors: remove duplicate base class members, update renamed APIs

- Remove duplicate OutputLog, WriteInputAsync, CreateTestTimeoutCts, etc. from
  ConsoleAppSamplesValidation (already in SamplesValidationBase)
- Update AddFanInEdge -> AddFanInBarrierEdge in workflow samples
- Update GetNewSessionAsync -> CreateSessionAsync in workflow samples
- Update SourceId -> ExecutorId (obsolete) in workflow samples

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix dotnet format issues: add UTF-8 BOM and remove unused using

- Add UTF-8 BOM to 20 .cs files across DurableTask, AzureFunctions,
  unit tests, and workflow samples
- Remove unnecessary using directive in 07_SubWorkflows/Executors.cs

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix typo PaymentProcesser -> PaymentProcessor and garbled arrows in README

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix GetExecutorName to handle agent names with underscores

Split on last underscore instead of first, and validate that the
suffix is a 32-char hex string (sanitized GUID) before stripping it.
This prevents truncation of agent names like 'my_agent' when the
executor ID is 'my_agent_<guid>'.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Align DurableTask.Client.AzureManaged to 1.19.1

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Bump DurableTask and Azure Functions extension package versions

- DurableTask.* packages: 1.19.1 -> 1.22.0
- Functions.Worker.Extensions.DurableTask: 1.13.1 -> 1.16.0
- Functions.Worker.Extensions.DurableTask.AzureManaged: 1.0.1 -> 1.5.0 (telemetry bug fix)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Bump DurableTask SDK packages to 1.22.0

- DurableTask.Client: 1.19.1 -> 1.22.0
- DurableTask.Client.AzureManaged: 1.19.1 -> 1.22.0
- DurableTask.Worker: 1.19.1 -> 1.22.0
- DurableTask.Worker.AzureManaged: 1.19.1 -> 1.22.0
- Azure Functions extensions kept at original versions (1.13.1/1.0.1) due to
  host-side DurableTask.Core 3.7.0 incompatibility with newer extensions

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Update Microsoft.Azure.Functions.Worker.Extensions.DurableTask to "1.16.0"

* Add the local.settings.json files to the sample which were previously ignored. This aligns with our other samples.

* Increase timeout for tests as CI has them failing transiently.

* increaset timeout value for azure functions integration tests.

* Add YieldsOutput(string) to workflow shared state sample executors

ValidateOrder and EnrichOrder call YieldOutputAsync with string messages,
but only their TOutput (OrderDetails) was in the allowed yield types.
This caused TargetInvocationException in the WorkflowSharedState sample
validation integration test.

* Downgrade the durable packages to 1.18.0

* Downgrading Worker.Extensions.DurableTask to 1.12.1

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-16 23:00:50 +00:00
0fdcfd0f4c Python: preserve A2A message context_id (#4686)
* Python: forward A2A context_id

* Avoid duplicating A2A context ids

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-16 21:41:31 +00:00
Giles OdigweandGitHub 414496dda7 fix: Azure Redis sample missing session for history persistence (#4692) 2026-03-16 21:34:21 +00:00
55011b7258 Python: Fix _deduplicate_messages catch-all branch dropping valid repeated messages (#4716)
* Fix _deduplicate_messages catch-all branch dropping valid repeated messages (#4682)

Remove the catch-all dedup branch that used (role, hash(content_str)) as a
dedup key. This incorrectly treated any two messages with the same role and
identical content as duplicates, dropping valid repeated messages (e.g., a
user saying 'yes' to confirm two separate things).

The tool-specific dedup branches (tool results by call_id, assistant tool
calls by call_id tuple) remain unchanged as they correctly identify true
protocol-level duplicates.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Address review: consecutive-duplicate detection for non-tool messages (#4682)

- Replace blanket dedup removal with consecutive-duplicate detection:
  only skip a message if the immediately preceding message has the same
  role and content, preserving protection against upstream replays while
  allowing identical messages at different conversation points.
- Strengthen test assertions to verify message identity and order, not
  just list length.
- Add tests for consecutive duplicate skipping, non-consecutive
  preservation, and messages with contents=None.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Apply pre-commit auto-fixes

* Use message_id for deduplication instead of content hashing

Deduplicate general messages by message_id when available, replacing
the consecutive-duplicate content check. Two messages with the same id
are definitively the same message (upstream replay), while identical
content with distinct ids (e.g. repeated "yes" confirmations) is
preserved. Messages without a message_id are always kept.

* Fix message_id dedup: truthy check, content-hash fallback, log safety

- Use truthy check (`if msg.message_id`) instead of `is not None` so
  empty-string IDs fall through to content-hash dedup rather than
  collapsing unrelated messages.
- Add content-hash fallback for messages without message_id, preventing
  false negatives from integrations that don't set IDs.
- Remove raw message_id from log format string (addresses log-injection
  surface with control characters).
- Add tests for empty-string message_id edge cases.
- Update existing tests to reflect content-hash dedup behavior.

Fixes #4682

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-16 17:47:33 +00:00
CopilotGitHubcrickmancopilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
bf0af178bd .NET - Fix flaky workflows test (#4700)
* Initial plan

* Fix flaky test: initialize creationTime 1 second in the past

Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: crickman <66376200+crickman@users.noreply.github.com>
2026-03-16 17:33:08 +00:00
297 changed files with 20987 additions and 1800 deletions
+1
View File
@@ -3,6 +3,7 @@
"image": "mcr.microsoft.com/devcontainers/dotnet",
"features": {
"ghcr.io/devcontainers/features/azure-cli:1.2.9": {},
"ghcr.io/devcontainers/features/docker-in-docker:2": {},
"ghcr.io/devcontainers/features/github-cli:1": {
"version": "2"
},
+207
View File
@@ -0,0 +1,207 @@
# Copyright (c) Microsoft. All rights reserved.
"""Scan open issues and PRs for stale follow-ups from external authors.
If a team member commented and the external author hasn't replied within
DAYS_THRESHOLD days, post a reminder comment and add the 'needs-info' label.
"""
from __future__ import annotations
import os
import sys
import time
from datetime import datetime, timezone
from github import Auth, Github, GithubException
from github.Issue import Issue
from github.IssueComment import IssueComment
PING_COMMENT = (
"@{author}, friendly reminder — this issue is waiting on your response. "
"Please share any updates when you get a chance. (This is an automated message.)"
)
LABEL = "needs-info"
def get_team_members(g: Github, org: str, team_slug: str) -> set[str]:
"""Fetch active team member usernames."""
try:
org_obj = g.get_organization(org)
team = org_obj.get_team_by_slug(team_slug)
return {m.login for m in team.get_members()}
except GithubException as exc:
if exc.status in (403, 404):
print(
f"ERROR: Failed to fetch team members for {org}/{team_slug} "
f"(HTTP {exc.status}). Check that the token has the 'read:org' "
f"scope and that the team slug '{team_slug}' is correct."
)
else:
print(f"ERROR: Failed to fetch team members for {org}/{team_slug}: {exc}")
sys.exit(1)
except Exception as exc:
print(f"ERROR: Failed to fetch team members for {org}/{team_slug}: {exc}")
sys.exit(1)
def find_last_team_comment(
comments: list[IssueComment], team_members: set[str]
) -> IssueComment | None:
"""Return the most recent comment from a team member, or None."""
for comment in reversed(comments):
if comment.user and comment.user.login in team_members:
return comment
return None
def author_replied_after(
comments: list[IssueComment], author: str, after: datetime
) -> bool:
"""Check if the issue author commented after the given timestamp."""
for comment in comments:
if (
comment.user
and comment.user.login == author
and comment.created_at > after
):
return True
return False
def should_ping(
issue: Issue,
team_members: set[str],
days_threshold: int,
now: datetime,
) -> bool:
"""Determine whether this issue/PR should be pinged."""
author = issue.user.login
# Skip if author is a team member
if author in team_members:
return False
# Skip if already labeled
if any(label.name == LABEL for label in issue.labels):
return False
# Skip if no comments at all
if issue.comments == 0:
return False
# Fetch comments once for both lookups
comments = list(issue.get_comments())
# Find last team member comment
last_team_comment = find_last_team_comment(comments, team_members)
if last_team_comment is None:
return False
# Skip if author replied after the last team comment
if author_replied_after(comments, author, last_team_comment.created_at):
return False
# Check if enough days have passed
days_since = (now - last_team_comment.created_at.astimezone(timezone.utc)).days
if days_since < days_threshold:
return False
return True
def ping(issue: Issue, dry_run: bool) -> bool:
"""Post a reminder comment and add the needs-info label. Returns True on success."""
author = issue.user.login
kind = "PR" if issue.pull_request else "Issue"
if dry_run:
print(f" [DRY RUN] Would ping {kind} #{issue.number} (@{author})")
return True
max_retries = 3
commented = False
labeled = False
for attempt in range(1, max_retries + 1):
try:
if not commented:
issue.create_comment(PING_COMMENT.format(author=author))
commented = True
if not labeled:
issue.add_to_labels(LABEL)
labeled = True
print(f" Pinged {kind} #{issue.number} (@{author})")
return True
except Exception as exc:
if attempt < max_retries:
wait = 2 ** attempt # 2s, 4s
print(f" WARN: Attempt {attempt}/{max_retries} failed for {kind} #{issue.number}: {exc}. Retrying in {wait}s...")
time.sleep(wait)
else:
print(f" ERROR: Failed to ping {kind} #{issue.number} after {max_retries} attempts: {exc}")
return False
def main() -> None:
token = os.environ.get("GITHUB_TOKEN")
if not token:
print("ERROR: GITHUB_TOKEN environment variable is required")
sys.exit(1)
repository = os.environ.get("GITHUB_REPOSITORY")
if not repository:
print("ERROR: GITHUB_REPOSITORY environment variable is required")
sys.exit(1)
team_slug = os.environ.get("TEAM_SLUG")
if not team_slug:
print("ERROR: TEAM_SLUG environment variable is required")
sys.exit(1)
days_threshold_raw = os.environ.get("DAYS_THRESHOLD", "4")
try:
days_threshold = int(days_threshold_raw)
except ValueError:
print(f"ERROR: DAYS_THRESHOLD must be a numeric value, got '{days_threshold_raw}'")
sys.exit(1)
dry_run = os.environ.get("DRY_RUN", "false").lower() == "true"
org = repository.split("/")[0]
if dry_run:
print("Running in DRY RUN mode — no comments or labels will be applied.\n")
g = Github(auth=Auth.Token(token))
repo = g.get_repo(repository)
print(f"Fetching team members for {org}/{team_slug}...")
team_members = get_team_members(g, org, team_slug)
print(f"Found {len(team_members)} team members.\n")
now = datetime.now(timezone.utc)
pinged = []
failed = []
scanned = 0
print(f"Scanning open issues and PRs (threshold: {days_threshold} days)...\n")
for issue in repo.get_issues(state="open"):
scanned += 1
if should_ping(issue, team_members, days_threshold, now):
if ping(issue, dry_run):
pinged.append(issue.number)
else:
failed.append(issue.number)
print(f"\nDone. Scanned {scanned} items, pinged {len(pinged)}, failed {len(failed)}.")
if pinged:
print(f"Pinged: {', '.join(f'#{n}' for n in pinged)}")
if failed:
print(f"Failed: {', '.join(f'#{n}' for n in failed)}")
sys.exit(1)
if __name__ == "__main__":
main()
+293
View File
@@ -0,0 +1,293 @@
# Copyright (c) Microsoft. All rights reserved.
"""Tests for stale_issue_pr_ping.py."""
from __future__ import annotations
import os
import sys
from datetime import datetime, timezone, timedelta
from unittest.mock import MagicMock, patch
import pytest
# Ensure the script directory is importable
sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "scripts"))
from stale_issue_pr_ping import (
LABEL,
PING_COMMENT,
author_replied_after,
find_last_team_comment,
get_team_members,
main,
ping,
should_ping,
)
TEAM = {"alice", "bob"}
NOW = datetime(2026, 3, 15, 12, 0, 0, tzinfo=timezone.utc)
# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
def _make_comment(login: str | None, created_at: datetime) -> MagicMock:
"""Create a mock IssueComment."""
c = MagicMock()
if login is None:
c.user = None
else:
c.user = MagicMock()
c.user.login = login
c.created_at = created_at
return c
def _make_label(name: str) -> MagicMock:
lbl = MagicMock()
lbl.name = name
return lbl
def _make_issue(
author: str = "external",
labels: list[str] | None = None,
comment_count: int = 1,
comments: list[MagicMock] | None = None,
pull_request: bool = False,
number: int = 42,
) -> MagicMock:
issue = MagicMock()
issue.user = MagicMock()
issue.user.login = author
issue.number = number
issue.labels = [_make_label(n) for n in (labels or [])]
issue.comments = comment_count
issue.pull_request = MagicMock() if pull_request else None
if comments is not None:
issue.get_comments.return_value = comments
return issue
# ---------------------------------------------------------------------------
# find_last_team_comment
# ---------------------------------------------------------------------------
class TestFindLastTeamComment:
def test_returns_last_team_comment(self):
c1 = _make_comment("alice", datetime(2026, 3, 1, tzinfo=timezone.utc))
c2 = _make_comment("external", datetime(2026, 3, 2, tzinfo=timezone.utc))
c3 = _make_comment("bob", datetime(2026, 3, 3, tzinfo=timezone.utc))
assert find_last_team_comment([c1, c2, c3], TEAM) is c3
def test_returns_none_when_no_team_comments(self):
c1 = _make_comment("external", datetime(2026, 3, 1, tzinfo=timezone.utc))
assert find_last_team_comment([c1], TEAM) is None
def test_returns_none_for_empty_list(self):
assert find_last_team_comment([], TEAM) is None
def test_skips_deleted_user(self):
c1 = _make_comment(None, datetime(2026, 3, 1, tzinfo=timezone.utc))
c2 = _make_comment("alice", datetime(2026, 3, 2, tzinfo=timezone.utc))
assert find_last_team_comment([c1, c2], TEAM) is c2
def test_only_deleted_users(self):
c1 = _make_comment(None, datetime(2026, 3, 1, tzinfo=timezone.utc))
assert find_last_team_comment([c1], TEAM) is None
# ---------------------------------------------------------------------------
# author_replied_after
# ---------------------------------------------------------------------------
class TestAuthorRepliedAfter:
def test_author_replied(self):
after = datetime(2026, 3, 1, tzinfo=timezone.utc)
c1 = _make_comment("external", datetime(2026, 3, 2, tzinfo=timezone.utc))
assert author_replied_after([c1], "external", after) is True
def test_author_not_replied(self):
after = datetime(2026, 3, 5, tzinfo=timezone.utc)
c1 = _make_comment("external", datetime(2026, 3, 2, tzinfo=timezone.utc))
assert author_replied_after([c1], "external", after) is False
def test_different_user_replied(self):
after = datetime(2026, 3, 1, tzinfo=timezone.utc)
c1 = _make_comment("someone_else", datetime(2026, 3, 2, tzinfo=timezone.utc))
assert author_replied_after([c1], "external", after) is False
def test_deleted_user_comment(self):
after = datetime(2026, 3, 1, tzinfo=timezone.utc)
c1 = _make_comment(None, datetime(2026, 3, 2, tzinfo=timezone.utc))
assert author_replied_after([c1], "external", after) is False
# ---------------------------------------------------------------------------
# should_ping
# ---------------------------------------------------------------------------
class TestShouldPing:
def test_should_ping_stale_issue(self):
team_comment = _make_comment("alice", NOW - timedelta(days=5))
issue = _make_issue(comments=[team_comment], comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is True
def test_skip_team_member_author(self):
issue = _make_issue(author="alice", comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_skip_already_labeled(self):
issue = _make_issue(labels=[LABEL], comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_skip_no_comments(self):
issue = _make_issue(comment_count=0)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_skip_no_team_comment(self):
c = _make_comment("external", NOW - timedelta(days=5))
issue = _make_issue(comments=[c], comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_skip_author_replied(self):
team_c = _make_comment("alice", NOW - timedelta(days=5))
author_c = _make_comment("external", NOW - timedelta(days=3))
issue = _make_issue(comments=[team_c, author_c], comment_count=2)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_skip_not_enough_days(self):
team_comment = _make_comment("alice", NOW - timedelta(days=2))
issue = _make_issue(comments=[team_comment], comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is False
def test_aware_datetime_handled(self):
"""Timezone-aware datetimes should not be mangled by astimezone."""
aware_dt = (NOW - timedelta(days=5)).replace(tzinfo=timezone.utc)
team_comment = _make_comment("alice", aware_dt)
issue = _make_issue(comments=[team_comment], comment_count=1)
assert should_ping(issue, TEAM, 4, NOW) is True
def test_naive_datetime_handled(self):
"""Naive datetimes (pre-PyGithub 2.x) should be handled by astimezone."""
naive_dt = (NOW - timedelta(days=5)).replace(tzinfo=None)
team_comment = _make_comment("alice", naive_dt)
issue = _make_issue(comments=[team_comment], comment_count=1)
# astimezone on naive datetime treats it as local time; just verify no crash
should_ping(issue, TEAM, 4, NOW)
# ---------------------------------------------------------------------------
# ping
# ---------------------------------------------------------------------------
class TestPing:
def test_dry_run(self, capsys):
issue = _make_issue()
assert ping(issue, dry_run=True) is True
issue.create_comment.assert_not_called()
assert "DRY RUN" in capsys.readouterr().out
def test_success(self, capsys):
issue = _make_issue()
assert ping(issue, dry_run=False) is True
issue.create_comment.assert_called_once()
issue.add_to_labels.assert_called_once_with(LABEL)
@patch("stale_issue_pr_ping.time.sleep")
def test_retry_on_failure(self, mock_sleep):
issue = _make_issue()
issue.create_comment.side_effect = [Exception("net error"), None]
assert ping(issue, dry_run=False) is True
assert issue.create_comment.call_count == 2
mock_sleep.assert_called_once()
@patch("stale_issue_pr_ping.time.sleep")
def test_idempotent_retry_skips_comment_on_label_failure(self, mock_sleep):
"""If create_comment succeeds but add_to_labels fails, retry should not re-comment."""
issue = _make_issue()
issue.add_to_labels.side_effect = [Exception("label error"), None]
assert ping(issue, dry_run=False) is True
# Comment should only be created once even though there were 2 attempts
assert issue.create_comment.call_count == 1
assert issue.add_to_labels.call_count == 2
@patch("stale_issue_pr_ping.time.sleep")
def test_all_retries_fail(self, mock_sleep):
issue = _make_issue()
issue.create_comment.side_effect = Exception("permanent error")
assert ping(issue, dry_run=False) is False
assert issue.create_comment.call_count == 3
# ---------------------------------------------------------------------------
# get_team_members
# ---------------------------------------------------------------------------
class TestGetTeamMembers:
def test_success(self):
g = MagicMock()
member = MagicMock()
member.login = "alice"
g.get_organization.return_value.get_team_by_slug.return_value.get_members.return_value = [member]
assert get_team_members(g, "org", "my-team") == {"alice"}
def test_403_error_message(self, capsys):
from github import GithubException
g = MagicMock()
g.get_organization.return_value.get_team_by_slug.side_effect = GithubException(
403, {"message": "Forbidden"}, None
)
with pytest.raises(SystemExit):
get_team_members(g, "org", "my-team")
out = capsys.readouterr().out
assert "read:org" in out
assert "403" in out
def test_404_error_message(self, capsys):
from github import GithubException
g = MagicMock()
g.get_organization.return_value.get_team_by_slug.side_effect = GithubException(
404, {"message": "Not Found"}, None
)
with pytest.raises(SystemExit):
get_team_members(g, "org", "bad-slug")
out = capsys.readouterr().out
assert "read:org" in out
assert "bad-slug" in out
def test_generic_error(self, capsys):
g = MagicMock()
g.get_organization.side_effect = RuntimeError("boom")
with pytest.raises(SystemExit):
get_team_members(g, "org", "team")
# ---------------------------------------------------------------------------
# main env var validation
# ---------------------------------------------------------------------------
class TestMain:
@patch.dict(os.environ, {
"GITHUB_TOKEN": "tok",
"GITHUB_REPOSITORY": "org/repo",
"TEAM_SLUG": "my-team",
"DAYS_THRESHOLD": "abc",
}, clear=True)
def test_invalid_days_threshold(self, capsys):
with pytest.raises(SystemExit):
main()
assert "numeric" in capsys.readouterr().out
@patch.dict(os.environ, {
"GITHUB_TOKEN": "tok",
"GITHUB_REPOSITORY": "org/repo",
}, clear=True)
def test_missing_team_slug(self, capsys):
with pytest.raises(SystemExit):
main()
assert "TEAM_SLUG" in capsys.readouterr().out
+4 -4
View File
@@ -85,7 +85,7 @@ jobs:
workflow-samples
- name: Setup dotnet
uses: actions/setup-dotnet@v5.1.0
uses: actions/setup-dotnet@v5.2.0
with:
global-json-file: ${{ github.workspace }}/dotnet/global.json
- name: Build dotnet solutions
@@ -165,7 +165,7 @@ jobs:
echo "COSMOSDB_EMULATOR_AVAILABLE=true" >> $env:GITHUB_ENV
- name: Setup dotnet
uses: actions/setup-dotnet@v5.1.0
uses: actions/setup-dotnet@v5.2.0
with:
global-json-file: ${{ github.workspace }}/dotnet/global.json
@@ -281,7 +281,7 @@ jobs:
# Generate test reports and check coverage
- name: Generate test reports
if: matrix.targetFramework == env.COVERAGE_FRAMEWORK
uses: danielpalme/ReportGenerator-GitHub-Action@5.5.1
uses: danielpalme/ReportGenerator-GitHub-Action@5.5.3
with:
reports: "./TestResults/Coverage/**/*.cobertura.xml"
targetdir: "./TestResults/Reports"
@@ -289,7 +289,7 @@ jobs:
- name: Upload coverage report artifact
if: matrix.targetFramework == env.COVERAGE_FRAMEWORK
uses: actions/upload-artifact@v6
uses: actions/upload-artifact@v7
with:
name: CoverageReport-${{ matrix.os }}-${{ matrix.targetFramework }}-${{ matrix.configuration }} # Artifact name
path: ./TestResults/Reports # Directory containing files to upload
@@ -50,7 +50,7 @@ jobs:
echo "COSMOS_EMULATOR_AVAILABLE=true" >> $env:GITHUB_ENV
- name: Setup dotnet
uses: actions/setup-dotnet@v5.1.0
uses: actions/setup-dotnet@v5.2.0
with:
global-json-file: ${{ github.workspace }}/dotnet/global.json
+4 -6
View File
@@ -75,7 +75,7 @@ jobs:
os: ${{ runner.os }}
env:
UV_CACHE_DIR: /tmp/.uv-cache
- name: Run fmt, lint, pyright in parallel across packages
- name: Run syntax and pyright across packages
run: uv run poe check-packages
samples-markdown:
@@ -104,10 +104,8 @@ jobs:
os: ${{ runner.os }}
env:
UV_CACHE_DIR: /tmp/.uv-cache
- name: Run samples lint
run: uv run poe samples-lint
- name: Run samples syntax check
run: uv run poe samples-syntax
- name: Run samples checks
run: uv run poe check -S
- name: Run markdown code lint
run: uv run poe markdown-code-lint
@@ -140,4 +138,4 @@ jobs:
- name: Run Mypy
env:
GITHUB_BASE_REF: ${{ github.event.pull_request.base.ref || github.base_ref || 'main' }}
run: uv run poe ci-mypy
run: uv run python scripts/workspace_poe_tasks.py ci-mypy
@@ -38,13 +38,13 @@ jobs:
id: validate_ranges
# Keep workflow running so we can still publish diagnostics from this run.
continue-on-error: true
run: uv run poe validate-dependency-bounds-project --mode upper --project "*"
run: uv run poe validate-dependency-bounds-project --mode upper --package "*"
working-directory: ./python
- name: Upload dependency range report
# Always publish the report so failures are inspectable even when validation fails.
if: always()
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
with:
name: dependency-range-results
path: python/scripts/dependencies/dependency-range-results.json
@@ -203,7 +203,7 @@ jobs:
cat > "${PR_BODY_FILE}" <<'EOF'
This PR was generated by the dependency range validation workflow.
- Ran `uv run poe validate-dependency-bounds-project --mode upper --project "*"`
- Ran `uv run poe validate-dependency-bounds-project --mode upper --package "*"`
- Updated package dependency bounds
- Refreshed `python/uv.lock` with `uv lock --upgrade`
EOF
@@ -48,9 +48,8 @@ jobs:
os: ${{ runner.os }}
- name: Test with pytest (unit tests only)
run: >
uv run poe all-tests
uv run poe test -A
-m "not integration"
-n logical --dist worksteal
--timeout=120 --session-timeout=900 --timeout_method thread
--retries 2 --retry-delay 5
+1 -2
View File
@@ -100,9 +100,8 @@ jobs:
os: ${{ runner.os }}
- name: Test with pytest (unit tests only)
run: >
uv run poe all-tests
uv run poe test -A
-m "not integration"
-n logical --dist worksteal
--timeout=120 --session-timeout=900 --timeout_method thread
--retries 2 --retry-delay 5
working-directory: ./python
@@ -46,7 +46,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir 01-get-started --save-report --report-name 01-get-started
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-01-get-started
@@ -89,7 +89,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir 02-agents --save-report --report-name 02-agents
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-02-agents
@@ -126,7 +126,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir 03-workflows --save-report --report-name 03-workflows
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-03-workflows
@@ -165,7 +165,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir 04-hosting --save-report --report-name 04-hosting
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-04-hosting
@@ -209,7 +209,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir 05-end-to-end --save-report --report-name 05-end-to-end
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-05-end-to-end
@@ -249,7 +249,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir autogen-migration --save-report --report-name autogen-migration
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-autogen-migration
@@ -295,7 +295,7 @@ jobs:
cd scripts && uv run python -m sample_validation --subdir semantic-kernel-migration --save-report --report-name semantic-kernel-migration
- name: Upload validation report
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
if: always()
with:
name: validation-report-semantic-kernel-migration
@@ -46,7 +46,7 @@ jobs:
echo "PR_NUMBER=$PR_NUMBER" >> "$GITHUB_ENV"
- name: Pytest coverage comment
id: coverageComment
uses: MishaKav/pytest-coverage-comment@v1.2.0
uses: MishaKav/pytest-coverage-comment@v1.6.0
with:
github-token: ${{ secrets.GH_ACTIONS_PR_WRITE }}
issue-number: ${{ env.PR_NUMBER }}
+3 -3
View File
@@ -32,17 +32,17 @@ jobs:
id: python-setup
uses: ./.github/actions/python-setup
with:
python-version: ${{ matrix.python-version }}
python-version: ${{ env.UV_PYTHON }}
os: ${{ runner.os }}
env:
# Configure a constant location for the uv cache
UV_CACHE_DIR: /tmp/.uv-cache
- name: Run all tests with coverage report
run: uv run poe all-tests-cov --cov-report=xml:python-coverage.xml -q --junitxml=pytest.xml
run: uv run poe test -A -C --cov-report=xml:python-coverage.xml -q --junitxml=pytest.xml
- name: Check coverage threshold
run: python ${{ github.workspace }}/.github/workflows/python-check-coverage.py python-coverage.xml ${{ env.COVERAGE_THRESHOLD }}
- name: Upload coverage report
uses: actions/upload-artifact@v6
uses: actions/upload-artifact@v7
with:
path: |
python/python-coverage.xml
+1 -1
View File
@@ -40,7 +40,7 @@ jobs:
UV_CACHE_DIR: /tmp/.uv-cache
# Unit tests
- name: Run all tests
run: uv run poe all-tests ${{ matrix.python-version == '3.10' && '--ignore-glob=packages/github_copilot/**' || '' }}
run: uv run poe test -A
working-directory: ./python
# Surface failing tests
+49
View File
@@ -0,0 +1,49 @@
name: Stale issue and PR ping
on:
schedule:
- cron: '0 0 * * *' # Midnight UTC daily
workflow_dispatch:
inputs:
days_threshold:
description: 'Days of silence before pinging the author'
required: false
default: '4'
dry_run:
description: 'Log what would be pinged without taking action'
required: false
default: 'false'
type: choice
options:
- 'false'
- 'true'
concurrency:
group: stale-issue-pr-ping
cancel-in-progress: true
jobs:
ping_stale:
name: "Ping stale issues and PRs"
runs-on: ubuntu-latest
permissions:
contents: read
issues: write
pull-requests: write
steps:
- uses: actions/checkout@v6
- uses: actions/setup-python@v5
with:
python-version: '3.13'
- name: Install dependencies
run: pip install PyGithub==2.6.0
- name: Run stale issue/PR ping
run: python .github/scripts/stale_issue_pr_ping.py
env:
GITHUB_TOKEN: ${{ secrets.GH_ACTIONS_PR_WRITE }}
TEAM_SLUG: ${{ secrets.DEVELOPER_TEAM }}
DAYS_THRESHOLD: ${{ github.event.inputs.days_threshold || '4' }}
DRY_RUN: ${{ github.event.inputs.dry_run || 'false' }}
+2 -1
View File
@@ -127,7 +127,7 @@
<!-- Azure Functions -->
<PackageVersion Include="Microsoft.Azure.Functions.Worker" Version="2.50.0" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.ApplicationInsights" Version="2.50.0" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask" Version="1.11.0" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask" Version="1.12.1" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask.AzureManaged" Version="1.0.1" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.Extensions.Http" Version="3.3.0" />
<PackageVersion Include="Microsoft.Azure.Functions.Worker.Extensions.Http.AspNetCore" Version="2.1.0" />
@@ -149,6 +149,7 @@
<!-- Symbols -->
<PackageVersion Include="Microsoft.SourceLink.GitHub" Version="8.0.0" />
<!-- Toolset -->
<PackageVersion Include="ReferenceTrimmer" Version="3.4.5" />
<PackageVersion Include="Microsoft.CodeAnalysis.Analyzers" Version="3.11.0" />
<PackageVersion Include="Microsoft.CodeAnalysis.CSharp" Version="4.14.0" />
<PackageVersion Include="Microsoft.CodeAnalysis.NetAnalyzers" Version="10.0.100" />
+20 -1
View File
@@ -61,6 +61,25 @@
<Folder Name="/Samples/02-agents/DeclarativeAgents/">
<Project Path="samples/02-agents/DeclarativeAgents/ChatClient/DeclarativeChatClientAgents.csproj" />
</Folder>
<Folder Name="/Samples/04-hosting/DurableWorkflows/" />
<Folder Name="/Samples/04-hosting/DurableWorkflows/ConsoleApps/">
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/01_SequentialWorkflow/01_SequentialWorkflow.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/02_ConcurrentWorkflow/02_ConcurrentWorkflow.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/03_ConditionalEdges/03_ConditionalEdges.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/04_WorkflowAndAgents/04_WorkflowAndAgents.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/05_WorkflowEvents/05_WorkflowEvents.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/06_WorkflowSharedState/06_WorkflowSharedState.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/07_SubWorkflows/07_SubWorkflows.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/ConsoleApps/08_WorkflowHITL/08_WorkflowHITL.csproj" />
</Folder>
<Folder Name="/Samples/04-hosting/DurableWorkflows/AzureFunctions/">
<Project Path="samples/04-hosting/DurableWorkflows/AzureFunctions/01_SequentialWorkflow/01_SequentialWorkflow.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/AzureFunctions/02_ConcurrentWorkflow/02_ConcurrentWorkflow.csproj" />
<Project Path="samples/04-hosting/DurableWorkflows/AzureFunctions/03_WorkflowHITL/03_WorkflowHITL.csproj" />
</Folder>
<Folder Name="/Samples/GettingStarted/">
<File Path="samples/GettingStarted/README.md" />
</Folder>
<Folder Name="/Samples/02-agents/AGUI/">
<File Path="samples/02-agents/AGUI/README.md" />
</Folder>
@@ -520,4 +539,4 @@
<Project Path="tests/Microsoft.Agents.AI.Workflows.Generators.UnitTests/Microsoft.Agents.AI.Workflows.Generators.UnitTests.csproj" />
<Project Path="tests/Microsoft.Agents.AI.Workflows.UnitTests/Microsoft.Agents.AI.Workflows.UnitTests.csproj" />
</Folder>
</Solution>
</Solution>
@@ -9,7 +9,7 @@
</PropertyGroup>
<ItemGroup>
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI\Microsoft.Agents.AI.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Abstractions\Microsoft.Agents.AI.Abstractions.csproj" />
</ItemGroup>
</Project>
@@ -12,13 +12,9 @@
<PackageReference Include="Azure.AI.OpenAI" />
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.Extensions.AI.OpenAI" />
<PackageReference Include="Microsoft.Agents.ObjectModel" />
<PackageReference Include="Microsoft.Agents.ObjectModel.Json" />
<PackageReference Include="Microsoft.Agents.ObjectModel.PowerFx" />
</ItemGroup>
<ItemGroup>
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Declarative\Microsoft.Agents.AI.Declarative.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
@@ -61,6 +61,12 @@ public static class Program
{
Console.WriteLine($"{outputEvent}");
}
if (evt is WorkflowErrorEvent errorEvent)
{
Console.WriteLine($"Workflow error: {errorEvent.Exception?.Message}");
Console.WriteLine($"Details: {errorEvent.Exception}");
}
}
}
}
@@ -175,7 +181,9 @@ internal sealed class FeedbackEvent(FeedbackResult feedbackResult) : WorkflowEve
/// <summary>
/// A custom executor that uses an AI agent to provide feedback on a slogan.
/// </summary>
internal sealed class FeedbackExecutor : Executor<SloganResult>
[SendsMessage(typeof(FeedbackResult))]
[YieldsOutput(typeof(string))]
internal sealed partial class FeedbackExecutor : Executor<SloganResult>
{
private readonly AIAgent _agent;
private AgentSession? _session;
@@ -14,7 +14,6 @@
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
<PackageReference Include="System.Net.ServerSentEvents" />
<PackageReference Include="Microsoft.Bcl.AsyncInterfaces" />
</ItemGroup>
<ItemGroup>
@@ -0,0 +1,42 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<AzureFunctionsVersion>v4</AzureFunctionsVersion>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<!-- The Functions build tools don't like namespaces that start with a number -->
<AssemblyName>SingleAgent</AssemblyName>
<RootNamespace>SingleAgent</RootNamespace>
</PropertyGroup>
<ItemGroup>
<FrameworkReference Include="Microsoft.AspNetCore.App" />
</ItemGroup>
<!-- Azure Functions packages -->
<ItemGroup>
<PackageReference Include="Microsoft.Azure.Functions.Worker" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask.AzureManaged" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.Http.AspNetCore" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Sdk" />
</ItemGroup>
<ItemGroup>
<PackageReference Include="Azure.AI.OpenAI" />
<PackageReference Include="Azure.Identity" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.Hosting.AzureFunctions" />
<PackageReference Include="Microsoft.Agents.AI.OpenAI" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.Hosting.AzureFunctions\Microsoft.Agents.AI.Hosting.AzureFunctions.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,215 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace SequentialWorkflow;
/// <summary>
/// Looks up an order by its ID and return an Order object.
/// </summary>
internal sealed class OrderLookup() : Executor<string, Order>("OrderLookup")
{
public override async ValueTask<Order> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] OrderLookup: Starting lookup for order '{message}'");
Console.ResetColor();
// Simulate database lookup with delay
await Task.Delay(TimeSpan.FromMicroseconds(100), cancellationToken);
Order order = new(
Id: message,
OrderDate: DateTime.UtcNow.AddDays(-1),
IsCancelled: false,
Customer: new Customer(Name: "Jerry", Email: "jerry@example.com"));
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine($"│ [Activity] OrderLookup: Found order '{message}' for customer '{order.Customer.Name}'");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return order;
}
}
/// <summary>
/// Cancels an order.
/// </summary>
internal sealed class OrderCancel() : Executor<Order, Order>("OrderCancel")
{
public override async ValueTask<Order> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] OrderCancel: Starting cancellation for order '{message.Id}'");
Console.ResetColor();
// Simulate a slow cancellation process (e.g., calling external payment system)
for (int i = 1; i <= 3; i++)
{
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine("│ [Activity] OrderCancel: Processing...");
Console.ResetColor();
}
Order cancelledOrder = message with { IsCancelled = true };
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($"│ [Activity] OrderCancel: ✓ Order '{cancelledOrder.Id}' has been cancelled");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return cancelledOrder;
}
}
/// <summary>
/// Sends a cancellation confirmation email to the customer.
/// </summary>
internal sealed class SendEmail() : Executor<Order, string>("SendEmail")
{
public override ValueTask<string> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] SendEmail: Sending email to '{message.Customer.Email}'...");
Console.ResetColor();
string result = $"Cancellation email sent for order {message.Id} to {message.Customer.Email}.";
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("│ [Activity] SendEmail: ✓ Email sent successfully!");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(result);
}
}
internal sealed record Order(string Id, DateTime OrderDate, bool IsCancelled, Customer Customer);
internal sealed record Customer(string Name, string Email);
/// <summary>
/// Represents a batch cancellation request with multiple order IDs and a reason.
/// This demonstrates using a complex typed object as workflow input.
/// </summary>
#pragma warning disable CA1812 // Instantiated via JSON deserialization at runtime
internal sealed record BatchCancelRequest(string[] OrderIds, string Reason, bool NotifyCustomers);
#pragma warning restore CA1812
/// <summary>
/// Represents the result of processing a batch cancellation.
/// </summary>
internal sealed record BatchCancelResult(int TotalOrders, int CancelledCount, string Reason);
/// <summary>
/// Generates a status report for an order.
/// </summary>
internal sealed class StatusReport() : Executor<Order, string>("StatusReport")
{
public override ValueTask<string> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] StatusReport: Generating report for order '{message.Id}'");
Console.ResetColor();
string status = message.IsCancelled ? "Cancelled" : "Active";
string result = $"Order {message.Id} for {message.Customer.Name}: Status={status}, Date={message.OrderDate:yyyy-MM-dd}";
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine($"│ [Activity] StatusReport: ✓ {result}");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(result);
}
}
/// <summary>
/// Processes a batch cancellation request. Accepts a complex <see cref="BatchCancelRequest"/> object
/// as input, demonstrating how workflows can receive structured JSON input.
/// </summary>
internal sealed class BatchCancelProcessor() : Executor<BatchCancelRequest, BatchCancelResult>("BatchCancelProcessor")
{
public override async ValueTask<BatchCancelResult> HandleAsync(
BatchCancelRequest message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] BatchCancelProcessor: Processing {message.OrderIds.Length} orders");
Console.WriteLine($"│ [Activity] BatchCancelProcessor: Reason: {message.Reason}");
Console.WriteLine($"│ [Activity] BatchCancelProcessor: Notify customers: {message.NotifyCustomers}");
Console.ResetColor();
// Simulate processing each order
int cancelledCount = 0;
foreach (string orderId in message.OrderIds)
{
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
cancelledCount++;
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($"│ [Activity] BatchCancelProcessor: ✓ Cancelled order '{orderId}'");
Console.ResetColor();
}
BatchCancelResult result = new(message.OrderIds.Length, cancelledCount, message.Reason);
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($"│ [Activity] BatchCancelProcessor: ✓ Batch complete: {cancelledCount}/{message.OrderIds.Length} cancelled");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return result;
}
}
/// <summary>
/// Generates a summary of the batch cancellation.
/// </summary>
internal sealed class BatchCancelSummary() : Executor<BatchCancelResult, string>("BatchCancelSummary")
{
public override ValueTask<string> HandleAsync(
BatchCancelResult message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine("│ [Activity] BatchCancelSummary: Generating summary");
Console.ResetColor();
string result = $"Batch cancellation complete: {message.CancelledCount}/{message.TotalOrders} orders cancelled. Reason: {message.Reason}";
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine($"│ [Activity] BatchCancelSummary: ✓ {result}");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(result);
}
}
@@ -0,0 +1,52 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates three workflows that share executors.
// The CancelOrder workflow cancels an order and notifies the customer.
// The OrderStatus workflow looks up an order and generates a status report.
// The BatchCancelOrders workflow accepts a complex JSON input to cancel multiple orders.
// Both CancelOrder and OrderStatus reuse the same OrderLookup executor, demonstrating executor sharing.
using Microsoft.Agents.AI.Hosting.AzureFunctions;
using Microsoft.Agents.AI.Workflows;
using Microsoft.Azure.Functions.Worker.Builder;
using Microsoft.Extensions.Hosting;
using SequentialWorkflow;
// Define executors for all workflows
OrderLookup orderLookup = new();
OrderCancel orderCancel = new();
SendEmail sendEmail = new();
StatusReport statusReport = new();
BatchCancelProcessor batchCancelProcessor = new();
BatchCancelSummary batchCancelSummary = new();
// Build the CancelOrder workflow: OrderLookup -> OrderCancel -> SendEmail
Workflow cancelOrder = new WorkflowBuilder(orderLookup)
.WithName("CancelOrder")
.WithDescription("Cancel an order and notify the customer")
.AddEdge(orderLookup, orderCancel)
.AddEdge(orderCancel, sendEmail)
.Build();
// Build the OrderStatus workflow: OrderLookup -> StatusReport
// This workflow shares the OrderLookup executor with the CancelOrder workflow.
Workflow orderStatus = new WorkflowBuilder(orderLookup)
.WithName("OrderStatus")
.WithDescription("Look up an order and generate a status report")
.AddEdge(orderLookup, statusReport)
.Build();
// Build the BatchCancelOrders workflow: BatchCancelProcessor -> BatchCancelSummary
// This workflow demonstrates using a complex JSON object as the workflow input.
Workflow batchCancelOrders = new WorkflowBuilder(batchCancelProcessor)
.WithName("BatchCancelOrders")
.WithDescription("Cancel multiple orders in a batch using a complex JSON input")
.AddEdge(batchCancelProcessor, batchCancelSummary)
.Build();
using IHost app = FunctionsApplication
.CreateBuilder(args)
.ConfigureFunctionsWebApplication()
.ConfigureDurableWorkflows(workflows => workflows.AddWorkflows(cancelOrder, orderStatus, batchCancelOrders))
.Build();
app.Run();
@@ -0,0 +1,100 @@
# Sequential Workflow Sample
This sample demonstrates how to use the Microsoft Agent Framework to create an Azure Functions app that hosts durable workflows with sequential executor chains. It showcases two workflows that share a common executor, demonstrating executor reuse across workflows.
## Key Concepts Demonstrated
- Defining workflows with sequential executor chains using `WorkflowBuilder`
- Sharing executors across multiple workflows (the `OrderLookup` executor is used by both workflows)
- Registering workflows with the Function app using `ConfigureDurableWorkflows`
- Durable orchestration ensuring workflows survive process restarts and failures
- Starting workflows via HTTP requests
- Viewing workflow execution history and status in the Durable Task Scheduler (DTS) dashboard
## Workflows
This sample defines two workflows:
1. **CancelOrder**: `OrderLookup``OrderCancel``SendEmail` — Looks up an order, cancels it, and sends a confirmation email.
2. **OrderStatus**: `OrderLookup``StatusReport` — Looks up an order and generates a status report.
Both workflows share the `OrderLookup` executor, which is registered only once by the framework.
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for more information on how to configure the environment, including how to install and run common sample dependencies.
## Running the Sample
With the environment setup and function app running, you can test the sample by sending HTTP requests to the workflow endpoints.
You can use the `demo.http` file to trigger the workflows, or a command line tool like `curl` as shown below:
### Cancel an Order
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/CancelOrder/run \
-H "Content-Type: text/plain" \
-d "12345"
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/CancelOrder/run `
-ContentType text/plain `
-Body "12345"
```
The response will confirm the workflow orchestration has started:
```text
Workflow orchestration started for CancelOrder. Orchestration runId: abc123def456
```
> **Tip:** You can provide a custom run ID by appending a `runId` query parameter:
>
> ```bash
> curl -X POST "http://localhost:7071/api/workflows/CancelOrder/run?runId=my-order-123" \
> -H "Content-Type: text/plain" \
> -d "12345"
> ```
>
> If not provided, a unique run ID is auto-generated.
In the function app logs, you will see the sequential execution of each executor:
```text
│ [Activity] OrderLookup: Starting lookup for order '12345'
│ [Activity] OrderLookup: Found order '12345' for customer 'Jerry'
│ [Activity] OrderCancel: Starting cancellation for order '12345'
│ [Activity] OrderCancel: ✓ Order '12345' has been cancelled
│ [Activity] SendEmail: Sending email to 'jerry@example.com'...
│ [Activity] SendEmail: ✓ Email sent successfully!
```
### Get Order Status
```bash
curl -X POST http://localhost:7071/api/workflows/OrderStatus/run \
-H "Content-Type: text/plain" \
-d "12345"
```
The `OrderStatus` workflow reuses the same `OrderLookup` executor and then generates a status report:
```text
│ [Activity] OrderLookup: Starting lookup for order '12345'
│ [Activity] OrderLookup: Found order '12345' for customer 'Jerry'
│ [Activity] StatusReport: Generating report for order '12345'
│ [Activity] StatusReport: ✓ Order 12345 for Jerry: Status=Active, Date=2025-01-01
```
### Viewing Workflows in the DTS Dashboard
After running a workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to visualize the completed orchestration, inspect inputs/outputs for each step, and view execution history.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
@@ -0,0 +1,26 @@
# Default endpoint address for local testing
@authority=http://localhost:7071
### Cancel an order
POST {{authority}}/api/workflows/CancelOrder/run
Content-Type: text/plain
12345
### Cancel an order with a custom run ID
POST {{authority}}/api/workflows/CancelOrder/run?runId=my-custom-id-123
Content-Type: text/plain
99999
### Get order status (shares OrderLookup executor with CancelOrder)
POST {{authority}}/api/workflows/OrderStatus/run
Content-Type: text/plain
12345
### Batch cancel orders with a complex JSON input
POST {{authority}}/api/workflows/BatchCancelOrders/run
Content-Type: application/json
{"orderIds": ["1001", "1002", "1003"], "reason": "Customer requested cancellation", "notifyCustomers": true}
@@ -0,0 +1,20 @@
{
"version": "2.0",
"logging": {
"logLevel": {
"Microsoft.Agents.AI.DurableTask": "Information",
"Microsoft.Agents.AI.Hosting.AzureFunctions": "Information",
"DurableTask": "Information",
"Microsoft.DurableTask": "Information"
}
},
"extensions": {
"durableTask": {
"hubName": "default",
"storageProvider": {
"type": "AzureManaged",
"connectionStringName": "DURABLE_TASK_SCHEDULER_CONNECTION_STRING"
}
}
}
}
@@ -0,0 +1,10 @@
{
"IsEncrypted": false,
"Values": {
"FUNCTIONS_WORKER_RUNTIME": "dotnet-isolated",
"AzureWebJobsStorage": "UseDevelopmentStorage=true",
"DURABLE_TASK_SCHEDULER_CONNECTION_STRING": "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None",
"AZURE_OPENAI_ENDPOINT": "<AZURE_OPENAI_ENDPOINT>",
"AZURE_OPENAI_DEPLOYMENT_NAME": "<AZURE_OPENAI_DEPLOYMENT_NAME>"
}
}
@@ -0,0 +1,42 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<AzureFunctionsVersion>v4</AzureFunctionsVersion>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<!-- The Functions build tools don't like namespaces that start with a number -->
<AssemblyName>SingleAgent</AssemblyName>
<RootNamespace>SingleAgent</RootNamespace>
</PropertyGroup>
<ItemGroup>
<FrameworkReference Include="Microsoft.AspNetCore.App" />
</ItemGroup>
<!-- Azure Functions packages -->
<ItemGroup>
<PackageReference Include="Microsoft.Azure.Functions.Worker" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask.AzureManaged" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.Http.AspNetCore" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Sdk" />
</ItemGroup>
<ItemGroup>
<PackageReference Include="Azure.AI.OpenAI" />
<PackageReference Include="Azure.Identity" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.Hosting.AzureFunctions" />
<PackageReference Include="Microsoft.Agents.AI.OpenAI" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.Hosting.AzureFunctions\Microsoft.Agents.AI.Hosting.AzureFunctions.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,73 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowConcurrency;
/// <summary>
/// Parses and validates the incoming question before sending to AI agents.
/// </summary>
internal sealed class ParseQuestionExecutor() : Executor<string, string>("ParseQuestion")
{
public override ValueTask<string> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine("│ [ParseQuestion] Preparing question for AI agents...");
string formattedQuestion = message.Trim();
if (!formattedQuestion.EndsWith('?'))
{
formattedQuestion += "?";
}
Console.WriteLine($"│ [ParseQuestion] Question: \"{formattedQuestion}\"");
Console.WriteLine("│ [ParseQuestion] → Sending to Physicist and Chemist in PARALLEL...");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(formattedQuestion);
}
}
/// <summary>
/// Aggregates responses from all AI agents into a comprehensive answer.
/// This is the Fan-in point where parallel results are collected.
/// </summary>
internal sealed class AggregatorExecutor() : Executor<string[], string>("Aggregator")
{
public override ValueTask<string> HandleAsync(
string[] message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Aggregator] 📋 Received {message.Length} AI agent responses");
Console.WriteLine("│ [Aggregator] Combining into comprehensive answer...");
Console.WriteLine("│ [Aggregator] ✓ Aggregation complete!");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
string aggregatedResult = "═══════════════════════════════════════════════════════════════\n" +
" AI EXPERT PANEL RESPONSES\n" +
"═══════════════════════════════════════════════════════════════\n\n";
for (int i = 0; i < message.Length; i++)
{
string expertLabel = i == 0 ? "⚛️ PHYSICIST" : "🧪 CHEMIST";
aggregatedResult += $"{expertLabel}:\n{message[i]}\n\n";
}
aggregatedResult += "═══════════════════════════════════════════════════════════════\n" +
$"Summary: Received perspectives from {message.Length} AI experts.\n" +
"═══════════════════════════════════════════════════════════════";
return ValueTask.FromResult(aggregatedResult);
}
}
@@ -0,0 +1,45 @@
// Copyright (c) Microsoft. All rights reserved.
using Azure;
using Azure.AI.OpenAI;
using Azure.Identity;
using Microsoft.Agents.AI;
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.Hosting.AzureFunctions;
using Microsoft.Agents.AI.Workflows;
using Microsoft.Azure.Functions.Worker.Builder;
using Microsoft.Extensions.Hosting;
using OpenAI.Chat;
using WorkflowConcurrency;
string endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")
?? throw new InvalidOperationException("AZURE_OPENAI_ENDPOINT is not set.");
string deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT")
?? throw new InvalidOperationException("AZURE_OPENAI_DEPLOYMENT is not set.");
string? azureOpenAiKey = Environment.GetEnvironmentVariable("AZURE_OPENAI_KEY");
// Create Azure OpenAI client
AzureOpenAIClient openAiClient = !string.IsNullOrEmpty(azureOpenAiKey)
? new AzureOpenAIClient(new Uri(endpoint), new AzureKeyCredential(azureOpenAiKey))
: new AzureOpenAIClient(new Uri(endpoint), new AzureCliCredential());
ChatClient chatClient = openAiClient.GetChatClient(deploymentName);
// Define the 4 executors for the workflow
ParseQuestionExecutor parseQuestion = new();
AIAgent physicist = chatClient.AsAIAgent("You are a physics expert. Be concise (2-3 sentences).", "Physicist");
AIAgent chemist = chatClient.AsAIAgent("You are a chemistry expert. Be concise (2-3 sentences).", "Chemist");
AggregatorExecutor aggregator = new();
// Build workflow: ParseQuestion -> [Physicist, Chemist] (parallel) -> Aggregator
Workflow workflow = new WorkflowBuilder(parseQuestion)
.WithName("ExpertReview")
.AddFanOutEdge(parseQuestion, [physicist, chemist])
.AddFanInBarrierEdge([physicist, chemist], aggregator)
.Build();
using IHost app = FunctionsApplication
.CreateBuilder(args)
.ConfigureFunctionsWebApplication()
.ConfigureDurableWorkflows(workflows => workflows.AddWorkflows(workflow))
.Build();
app.Run();
@@ -0,0 +1,90 @@
# Concurrent Workflow Sample
This sample demonstrates how to use the Microsoft Agent Framework to create an Azure Functions app that orchestrates concurrent execution of multiple AI agents using the fan-out/fan-in pattern within a durable workflow.
## Key Concepts Demonstrated
- Defining workflows with fan-out/fan-in edges for parallel execution using `WorkflowBuilder`
- Mixing custom executors with AI agents in a single workflow
- Concurrent execution of multiple AI agents (physics and chemistry experts)
- Response aggregation from parallel branches into a unified result
- Durable orchestration with automatic checkpointing and resumption from failures
- Viewing workflow execution history and status in the Durable Task Scheduler (DTS) dashboard
## Workflow
This sample defines a single workflow:
**ExpertReview**: `ParseQuestion` → [`Physicist`, `Chemist`] (parallel) → `Aggregator`
1. **ParseQuestion** — A custom executor that validates and formats the incoming question.
2. **Physicist** and **Chemist** — AI agents that run concurrently, each providing an expert perspective.
3. **Aggregator** — A custom executor that combines the parallel responses into a comprehensive answer.
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for more information on how to configure the environment, including how to install and run common sample dependencies.
This sample requires Azure OpenAI. Set the following environment variables:
- `AZURE_OPENAI_ENDPOINT` — Your Azure OpenAI endpoint URL.
- `AZURE_OPENAI_DEPLOYMENT` — The name of your chat model deployment.
- `AZURE_OPENAI_KEY` (optional) — Your Azure OpenAI API key. If not set, Azure CLI credentials are used.
## Running the Sample
With the environment setup and function app running, you can test the sample by sending an HTTP request with a science question to the workflow endpoint.
You can use the `demo.http` file to trigger the workflow, or a command line tool like `curl` as shown below:
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/ExpertReview/run \
-H "Content-Type: text/plain" \
-d "What is temperature?"
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/ExpertReview/run `
-ContentType text/plain `
-Body "What is temperature?"
```
The response will confirm the workflow orchestration has started:
```text
Workflow orchestration started for ExpertReview. Orchestration runId: abc123def456
```
> **Tip:** You can provide a custom run ID by appending a `runId` query parameter:
>
> ```bash
> curl -X POST "http://localhost:7071/api/workflows/ExpertReview/run?runId=my-review-123" \
> -H "Content-Type: text/plain" \
> -d "What is temperature?"
> ```
>
> If not provided, a unique run ID is auto-generated.
In the function app logs, you will see the fan-out/fan-in execution pattern:
```text
│ [ParseQuestion] Preparing question for AI agents...
│ [ParseQuestion] Question: "What is temperature?"
│ [ParseQuestion] → Sending to Physicist and Chemist in PARALLEL...
│ [Aggregator] 📋 Received 2 AI agent responses
│ [Aggregator] Combining into comprehensive answer...
│ [Aggregator] ✓ Aggregation complete!
```
The Physicist and Chemist AI agents execute concurrently, and the Aggregator combines their responses into a formatted expert panel result.
### Viewing Workflows in the DTS Dashboard
After running a workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to visualize the completed orchestration, inspect inputs/outputs for each step, and view execution history.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
@@ -0,0 +1,14 @@
# Default endpoint address for local testing
@authority=http://localhost:7071
### Prompt the agent
POST {{authority}}/api/workflows/ExpertReview/run
Content-Type: text/plain
What is temperature?
### Start with a custom run ID
POST {{authority}}/api/workflows/ExpertReview/run?runId=my-review-123
Content-Type: text/plain
What is gravity?
@@ -0,0 +1,20 @@
{
"version": "2.0",
"logging": {
"logLevel": {
"Microsoft.Agents.AI.DurableTask": "Information",
"Microsoft.Agents.AI.Hosting.AzureFunctions": "Information",
"DurableTask": "Information",
"Microsoft.DurableTask": "Information"
}
},
"extensions": {
"durableTask": {
"hubName": "default",
"storageProvider": {
"type": "AzureManaged",
"connectionStringName": "DURABLE_TASK_SCHEDULER_CONNECTION_STRING"
}
}
}
}
@@ -0,0 +1,10 @@
{
"IsEncrypted": false,
"Values": {
"FUNCTIONS_WORKER_RUNTIME": "dotnet-isolated",
"AzureWebJobsStorage": "UseDevelopmentStorage=true",
"DURABLE_TASK_SCHEDULER_CONNECTION_STRING": "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None",
"AZURE_OPENAI_ENDPOINT": "<AZURE_OPENAI_ENDPOINT>",
"AZURE_OPENAI_DEPLOYMENT_NAME": "<AZURE_OPENAI_DEPLOYMENT_NAME>"
}
}
@@ -0,0 +1,43 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<AzureFunctionsVersion>v4</AzureFunctionsVersion>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<!-- The Functions build tools don't like namespaces that start with a number -->
<AssemblyName>WorkflowHITLFunctions</AssemblyName>
<RootNamespace>WorkflowHITLFunctions</RootNamespace>
</PropertyGroup>
<ItemGroup>
<FrameworkReference Include="Microsoft.AspNetCore.App" />
</ItemGroup>
<ItemGroup>
<None Include="local.settings.json" />
</ItemGroup>
<!-- Azure Functions packages -->
<ItemGroup>
<PackageReference Include="Microsoft.Azure.Functions.Worker" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.DurableTask.AzureManaged" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Extensions.Http.AspNetCore" />
<PackageReference Include="Microsoft.Azure.Functions.Worker.Sdk" />
</ItemGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.Hosting.AzureFunctions" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.Hosting.AzureFunctions\Microsoft.Agents.AI.Hosting.AzureFunctions.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,63 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowHITLFunctions;
/// <summary>Expense approval request passed to the RequestPort.</summary>
public record ApprovalRequest(string ExpenseId, decimal Amount, string EmployeeName);
/// <summary>Approval response received from the RequestPort.</summary>
public record ApprovalResponse(bool Approved, string? Comments);
/// <summary>Looks up expense details and creates an approval request.</summary>
internal sealed class CreateApprovalRequest() : Executor<string, ApprovalRequest>("RetrieveRequest")
{
public override ValueTask<ApprovalRequest> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// In a real scenario, this would look up expense details from a database
return new ValueTask<ApprovalRequest>(new ApprovalRequest(message, 1500.00m, "Jerry"));
}
}
/// <summary>Prepares the approval request for finance review after manager approval.</summary>
internal sealed class PrepareFinanceReview() : Executor<ApprovalResponse, ApprovalRequest>("PrepareFinanceReview")
{
public override ValueTask<ApprovalRequest> HandleAsync(
ApprovalResponse message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
if (!message.Approved)
{
throw new InvalidOperationException("Cannot proceed to finance review — manager denied the expense.");
}
// In a real scenario, this would retrieve the original expense details
return new ValueTask<ApprovalRequest>(new ApprovalRequest("EXP-2025-001", 1500.00m, "Jerry"));
}
}
/// <summary>Processes the expense reimbursement based on the parallel approval responses.</summary>
internal sealed class ExpenseReimburse() : Executor<ApprovalResponse[], string>("Reimburse")
{
public override async ValueTask<string> HandleAsync(
ApprovalResponse[] message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// Check that all parallel approvals passed
ApprovalResponse? denied = Array.Find(message, r => !r.Approved);
if (denied is not null)
{
return $"Expense reimbursement denied. Comments: {denied.Comments}";
}
// Simulate payment processing
await Task.Delay(1000, cancellationToken);
return $"Expense reimbursed at {DateTime.UtcNow:O}";
}
}
@@ -0,0 +1,51 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates a Human-in-the-Loop (HITL) workflow hosted in Azure Functions.
//
// ┌──────────────────────┐ ┌────────────────┐ ┌─────────────────────┐ ┌────────────────────┐
// │ CreateApprovalRequest│──►│ManagerApproval │──►│PrepareFinanceReview │──┬►│ BudgetApproval │──┐
// └──────────────────────┘ │ (RequestPort) │ └─────────────────────┘ │ │ (RequestPort) │ │
// └────────────────┘ │ └────────────────────┘ │ ┌─────────────────┐
// │ ├─►│ExpenseReimburse │
// │ ┌────────────────────┐ │ └─────────────────┘
// └►│ComplianceApproval │──┘
// │ (RequestPort) │
// └────────────────────┘
//
// The workflow pauses at three RequestPorts — one for the manager, then two in parallel for finance.
// After manager approval, BudgetApproval and ComplianceApproval run concurrently via fan-out/fan-in.
// The framework auto-generates three HTTP endpoints for each workflow:
// POST /api/workflows/{name}/run - Start the workflow
// GET /api/workflows/{name}/status/{id} - Check status and pending approvals
// POST /api/workflows/{name}/respond/{id} - Send approval response to resume
using Microsoft.Agents.AI.Hosting.AzureFunctions;
using Microsoft.Agents.AI.Workflows;
using Microsoft.Azure.Functions.Worker.Builder;
using Microsoft.Extensions.Hosting;
using WorkflowHITLFunctions;
// Define executors and RequestPorts for the three HITL pause points
CreateApprovalRequest createRequest = new();
RequestPort<ApprovalRequest, ApprovalResponse> managerApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("ManagerApproval");
PrepareFinanceReview prepareFinanceReview = new();
RequestPort<ApprovalRequest, ApprovalResponse> budgetApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("BudgetApproval");
RequestPort<ApprovalRequest, ApprovalResponse> complianceApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("ComplianceApproval");
ExpenseReimburse reimburse = new();
// Build the workflow: CreateApprovalRequest -> ManagerApproval -> PrepareFinanceReview -> [BudgetApproval AND ComplianceApproval] -> ExpenseReimburse
Workflow expenseApproval = new WorkflowBuilder(createRequest)
.WithName("ExpenseReimbursement")
.WithDescription("Expense reimbursement with manager and parallel finance approvals")
.AddEdge(createRequest, managerApproval)
.AddEdge(managerApproval, prepareFinanceReview)
.AddFanOutEdge(prepareFinanceReview, [budgetApproval, complianceApproval])
.AddFanInBarrierEdge([budgetApproval, complianceApproval], reimburse)
.Build();
using IHost app = FunctionsApplication
.CreateBuilder(args)
.ConfigureFunctionsWebApplication()
.ConfigureDurableWorkflows(workflows => workflows.AddWorkflow(expenseApproval, exposeStatusEndpoint: true))
.Build();
app.Run();
@@ -0,0 +1,266 @@
# Human-in-the-Loop (HITL) Workflow — Azure Functions
This sample demonstrates a durable workflow with Human-in-the-Loop support hosted in Azure Functions. The workflow pauses at three `RequestPort` nodes — one sequential manager approval, then two parallel finance approvals (budget and compliance) via fan-out/fan-in. Approval responses are sent via HTTP endpoints.
## Key Concepts Demonstrated
- Using multiple `RequestPort` nodes for sequential and parallel human-in-the-loop interactions in a durable workflow
- Fan-out/fan-in pattern for parallel approval steps
- Auto-generated HTTP endpoints for running workflows, checking status, and sending HITL responses
- Pausing orchestrations via `WaitForExternalEvent` and resuming via `RaiseEventAsync`
- Viewing inputs the workflow is waiting for via the status endpoint
## Workflow
This sample implements the following workflow:
```
┌──────────────────────┐ ┌────────────────┐ ┌─────────────────────┐ ┌────────────────────┐
│ CreateApprovalRequest│──►│ManagerApproval │──►│PrepareFinanceReview │──┬►│ BudgetApproval │──┐
└──────────────────────┘ │ (RequestPort) │ └─────────────────────┘ │ │ (RequestPort) │ │
└────────────────┘ │ └────────────────────┘ │ ┌─────────────────┐
│ ├─►│ExpenseReimburse │
│ ┌────────────────────┐ │ └─────────────────┘
└►│ComplianceApproval │──┘
│ (RequestPort) │
└────────────────────┘
```
## HTTP Endpoints
The framework auto-generates these endpoints for workflows with `RequestPort` nodes:
| Method | Endpoint | Description |
|--------|----------|-------------|
| POST | `/api/workflows/ExpenseReimbursement/run` | Start the workflow |
| GET | `/api/workflows/ExpenseReimbursement/status/{runId}` | Check status and inputs the workflow is waiting for |
| POST | `/api/workflows/ExpenseReimbursement/respond/{runId}` | Send approval response to resume |
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on how to configure the environment, including how to install and run the Durable Task Scheduler.
## Running the Sample
With the environment setup and function app running, you can test the sample by sending HTTP requests to the workflow endpoints.
You can use the `demo.http` file to trigger the workflow, or a command line tool like `curl` as shown below:
### Step 1: Start the Workflow
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/ExpenseReimbursement/run \
-H "Content-Type: text/plain" -d "EXP-2025-001"
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/ExpenseReimbursement/run `
-ContentType text/plain `
-Body "EXP-2025-001"
```
The response will confirm the workflow orchestration has started:
```text
Workflow orchestration started for ExpenseReimbursement. Orchestration runId: abc123def456
```
> [!TIP]
> You can provide a custom run ID by appending a `runId` query parameter:
>
> Bash (Linux/macOS/WSL):
>
> ```bash
> curl -X POST "http://localhost:7071/api/workflows/ExpenseReimbursement/run?runId=expense-001" \
> -H "Content-Type: text/plain" -d "EXP-2025-001"
> ```
>
> PowerShell:
>
> ```powershell
> Invoke-RestMethod -Method Post `
> -Uri "http://localhost:7071/api/workflows/ExpenseReimbursement/run?runId=expense-001" `
> -ContentType text/plain `
> -Body "EXP-2025-001"
> ```
>
> If not provided, a unique run ID is auto-generated.
### Step 2: Check Workflow Status
The workflow pauses at the `ManagerApproval` RequestPort. Query the status endpoint to see what input it is waiting for:
Bash (Linux/macOS/WSL):
```bash
curl http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
PowerShell:
```powershell
Invoke-RestMethod -Uri http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
```json
{
"runId": "{runId}",
"status": "Running",
"waitingForInput": [
{ "eventName": "ManagerApproval", "input": { "ExpenseId": "EXP-2025-001", "Amount": 1500.00, "EmployeeName": "Jerry" } }
]
}
```
> [!TIP]
> You can also verify this in the DTS dashboard at `http://localhost:8082`. Find the orchestration by its `runId` and you will see it is in a "Running" state, paused at a `WaitForExternalEvent` call for the `ManagerApproval` event.
### Step 3: Send Manager Approval Response
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} \
-H "Content-Type: application/json" \
-d '{"eventName": "ManagerApproval", "response": {"Approved": true, "Comments": "Approved by manager."}}'
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} `
-ContentType application/json `
-Body '{"eventName": "ManagerApproval", "response": {"Approved": true, "Comments": "Approved by manager."}}'
```
```json
{
"message": "Response sent to workflow.",
"runId": "{runId}",
"eventName": "ManagerApproval",
"validated": true
}
```
### Step 4: Check Workflow Status Again
The workflow now pauses at both the `BudgetApproval` and `ComplianceApproval` RequestPorts in parallel:
Bash (Linux/macOS/WSL):
```bash
curl http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
PowerShell:
```powershell
Invoke-RestMethod -Uri http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
```json
{
"runId": "{runId}",
"status": "Running",
"waitingForInput": [
{ "eventName": "BudgetApproval", "input": { "ExpenseId": "EXP-2025-001", "Amount": 1500.00, "EmployeeName": "Jerry" } },
{ "eventName": "ComplianceApproval", "input": { "ExpenseId": "EXP-2025-001", "Amount": 1500.00, "EmployeeName": "Jerry" } }
]
}
```
### Step 5a: Send Budget Approval Response
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} \
-H "Content-Type: application/json" \
-d '{"eventName": "BudgetApproval", "response": {"Approved": true, "Comments": "Budget approved."}}'
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} `
-ContentType application/json `
-Body '{"eventName": "BudgetApproval", "response": {"Approved": true, "Comments": "Budget approved."}}'
```
```json
{
"message": "Response sent to workflow.",
"runId": "{runId}",
"eventName": "BudgetApproval",
"validated": true
}
```
### Step 5b: Send Compliance Approval Response
Bash (Linux/macOS/WSL):
```bash
curl -X POST http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} \
-H "Content-Type: application/json" \
-d '{"eventName": "ComplianceApproval", "response": {"Approved": true, "Comments": "Compliance approved."}}'
```
PowerShell:
```powershell
Invoke-RestMethod -Method Post `
-Uri http://localhost:7071/api/workflows/ExpenseReimbursement/respond/{runId} `
-ContentType application/json `
-Body '{"eventName": "ComplianceApproval", "response": {"Approved": true, "Comments": "Compliance approved."}}'
```
```json
{
"message": "Response sent to workflow.",
"runId": "{runId}",
"eventName": "ComplianceApproval",
"validated": true
}
```
### Step 6: Check Final Status
After all approvals, the workflow completes and the expense is reimbursed:
Bash (Linux/macOS/WSL):
```bash
curl http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
PowerShell:
```powershell
Invoke-RestMethod -Uri http://localhost:7071/api/workflows/ExpenseReimbursement/status/{runId}
```
```json
{
"runId": "{runId}",
"status": "Completed",
"waitingForInput": null
}
```
### Viewing Workflows in the DTS Dashboard
After running a workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to visualize the orchestration and inspect its execution history.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
1. Open the dashboard and look for the orchestration instance matching the `runId` returned in Step 1 (e.g., `abc123def456` or your custom ID like `expense-001`).
2. Click into the instance to see the execution timeline, which shows each executor activity and the `WaitForExternalEvent` pauses where the workflow waited for human input — including the two parallel finance approvals.
3. Expand individual activity steps to inspect inputs and outputs — for example, the `ManagerApproval`, `BudgetApproval`, and `ComplianceApproval` external events will show the approval request sent and the response received.
@@ -0,0 +1,53 @@
# Default endpoint address for local testing
@authority=http://localhost:7071
### Step 1: Start the expense reimbursement workflow
POST {{authority}}/api/workflows/ExpenseReimbursement/run
Content-Type: text/plain
EXP-2025-001
### Step 1 (alternative): Start the workflow with a custom run ID
POST {{authority}}/api/workflows/ExpenseReimbursement/run?runId=expense-001
Content-Type: text/plain
EXP-2025-001
### Step 2: Check workflow status (replace {runId} with actual run ID from Step 1)
GET {{authority}}/api/workflows/ExpenseReimbursement/status/{runId}
### Step 3: Send manager approval (replace {runId} with actual run ID from Step 1)
POST {{authority}}/api/workflows/ExpenseReimbursement/respond/{runId}
Content-Type: application/json
{"eventName": "ManagerApproval", "response": {"Approved": true, "Comments": "Approved by manager."}}
### Step 3 (alternative): Deny the expense at manager level
POST {{authority}}/api/workflows/ExpenseReimbursement/respond/{runId}
Content-Type: application/json
{"eventName": "ManagerApproval", "response": {"Approved": false, "Comments": "Insufficient documentation. Please resubmit."}}
### Step 4: Check workflow status after manager approval (now waiting for parallel finance approvals)
GET {{authority}}/api/workflows/ExpenseReimbursement/status/{runId}
### Step 5a: Send budget approval (replace {runId} with actual run ID from Step 1)
POST {{authority}}/api/workflows/ExpenseReimbursement/respond/{runId}
Content-Type: application/json
{"eventName": "BudgetApproval", "response": {"Approved": true, "Comments": "Budget approved."}}
### Step 5b: Send compliance approval (replace {runId} with actual run ID from Step 1)
POST {{authority}}/api/workflows/ExpenseReimbursement/respond/{runId}
Content-Type: application/json
{"eventName": "ComplianceApproval", "response": {"Approved": true, "Comments": "Compliance approved."}}
### Step 5b (alternative): Deny the expense at compliance level
POST {{authority}}/api/workflows/ExpenseReimbursement/respond/{runId}
Content-Type: application/json
{"eventName": "ComplianceApproval", "response": {"Approved": false, "Comments": "Compliance requirements not met."}}
### Step 6: Check final workflow status after all approvals
GET {{authority}}/api/workflows/ExpenseReimbursement/status/{runId}
@@ -0,0 +1,20 @@
{
"version": "2.0",
"logging": {
"logLevel": {
"Microsoft.Agents.AI.DurableTask": "Information",
"Microsoft.Agents.AI.Hosting.AzureFunctions": "Information",
"DurableTask": "Information",
"Microsoft.DurableTask": "Information"
}
},
"extensions": {
"durableTask": {
"hubName": "default",
"storageProvider": {
"type": "AzureManaged",
"connectionStringName": "DURABLE_TASK_SCHEDULER_CONNECTION_STRING"
}
}
}
}
@@ -0,0 +1,10 @@
{
"IsEncrypted": false,
"Values": {
"FUNCTIONS_WORKER_RUNTIME": "dotnet-isolated",
"AzureWebJobsStorage": "UseDevelopmentStorage=true",
"DURABLE_TASK_SCHEDULER_CONNECTION_STRING": "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None",
"AZURE_OPENAI_ENDPOINT": "<AZURE_OPENAI_ENDPOINT>",
"AZURE_OPENAI_DEPLOYMENT_NAME": "<AZURE_OPENAI_DEPLOYMENT_NAME>"
}
}
@@ -0,0 +1,29 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>SequentialWorkflow</AssemblyName>
<RootNamespace>SequentialWorkflow</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,116 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace SequentialWorkflow;
/// <summary>
/// Represents a request to cancel an order.
/// </summary>
/// <param name="OrderId">The ID of the order to cancel.</param>
/// <param name="Reason">The reason for cancellation.</param>
internal sealed record OrderCancelRequest(string OrderId, string Reason);
/// <summary>
/// Looks up an order by its ID and return an Order object.
/// </summary>
internal sealed class OrderLookup() : Executor<OrderCancelRequest, Order>("OrderLookup")
{
public override async ValueTask<Order> HandleAsync(
OrderCancelRequest message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] OrderLookup: Starting lookup for order '{message.OrderId}'");
Console.WriteLine($"│ [Activity] OrderLookup: Cancellation reason: '{message.Reason}'");
Console.ResetColor();
// Simulate database lookup with delay
await Task.Delay(TimeSpan.FromMicroseconds(100), cancellationToken);
Order order = new(
Id: message.OrderId,
OrderDate: DateTime.UtcNow.AddDays(-1),
IsCancelled: false,
CancelReason: message.Reason,
Customer: new Customer(Name: "Jerry", Email: "jerry@example.com"));
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine($"│ [Activity] OrderLookup: Found order '{message.OrderId}' for customer '{order.Customer.Name}'");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return order;
}
}
/// <summary>
/// Cancels an order.
/// </summary>
internal sealed class OrderCancel() : Executor<Order, Order>("OrderCancel")
{
public override async ValueTask<Order> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// Log that this activity is executing (not replaying)
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] OrderCancel: Starting cancellation for order '{message.Id}'");
Console.ResetColor();
// Simulate a slow cancellation process (e.g., calling external payment system)
for (int i = 1; i <= 3; i++)
{
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine("│ [Activity] OrderCancel: Processing...");
Console.ResetColor();
}
Order cancelledOrder = message with { IsCancelled = true };
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($"│ [Activity] OrderCancel: ✓ Order '{cancelledOrder.Id}' has been cancelled");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return cancelledOrder;
}
}
/// <summary>
/// Sends a cancellation confirmation email to the customer.
/// </summary>
internal sealed class SendEmail() : Executor<Order, string>("SendEmail")
{
public override ValueTask<string> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Activity] SendEmail: Sending email to '{message.Customer.Email}'...");
Console.ResetColor();
string result = $"Cancellation email sent for order {message.Id} to {message.Customer.Email}.";
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("│ [Activity] SendEmail: ✓ Email sent successfully!");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(result);
}
}
internal sealed record Order(string Id, DateTime OrderDate, bool IsCancelled, string? CancelReason, Customer Customer);
internal sealed record Customer(string Name, string Email);
@@ -0,0 +1,93 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using SequentialWorkflow;
// Get DTS connection string from environment variable
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Define executors for the workflow
OrderLookup orderLookup = new();
OrderCancel orderCancel = new();
SendEmail sendEmail = new();
// Build the CancelOrder workflow: OrderLookup -> OrderCancel -> SendEmail
Workflow cancelOrder = new WorkflowBuilder(orderLookup)
.WithName("CancelOrder")
.WithDescription("Cancel an order and notify the customer")
.AddEdge(orderLookup, orderCancel)
.AddEdge(orderCancel, sendEmail)
.Build();
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
workflowOptions => workflowOptions.AddWorkflow(cancelOrder),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Durable Workflow Sample");
Console.WriteLine("Workflow: OrderLookup -> OrderCancel -> SendEmail");
Console.WriteLine();
Console.WriteLine("Enter an order ID (or 'exit'):");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
OrderCancelRequest request = new(OrderId: input, Reason: "Customer requested cancellation");
await StartNewWorkflowAsync(request, cancelOrder, workflowClient);
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
// Start a new workflow using IWorkflowClient with typed input
static async Task StartNewWorkflowAsync(OrderCancelRequest request, Workflow workflow, IWorkflowClient client)
{
Console.WriteLine($"Starting workflow for order '{request.OrderId}' (Reason: {request.Reason})...");
// RunAsync returns IWorkflowRun, cast to IAwaitableWorkflowRun for completion waiting
IAwaitableWorkflowRun run = (IAwaitableWorkflowRun)await client.RunAsync(workflow, request);
Console.WriteLine($"Run ID: {run.RunId}");
try
{
Console.WriteLine("Waiting for workflow to complete...");
string? result = await run.WaitForCompletionAsync<string>();
Console.WriteLine($"Workflow completed. {result}");
}
catch (InvalidOperationException ex)
{
Console.WriteLine($"Failed: {ex.Message}");
}
}
@@ -0,0 +1,83 @@
# Sequential Workflow Sample
This sample demonstrates how to run a sequential workflow as a durable orchestration from a console application using the Durable Task Framework. It showcases the **durability** aspect - if the process crashes mid-execution, the workflow automatically resumes without re-executing completed activities.
## Key Concepts Demonstrated
- Building a sequential workflow with the `WorkflowBuilder` API
- Using `ConfigureDurableWorkflows` to register workflows with dependency injection
- Running workflows with `IWorkflowClient`
- **Durability**: Automatic resume of interrupted workflows
- **Activity caching**: Completed activities are not re-executed on replay
## Overview
The sample implements an order cancellation workflow with three executors:
```
OrderLookup --> OrderCancel --> SendEmail
```
| Executor | Description |
|----------|-------------|
| OrderLookup | Looks up an order by ID |
| OrderCancel | Marks the order as cancelled |
| SendEmail | Sends a cancellation confirmation email |
## Durability Demonstration
The key feature of Durable Task Framework is **durability**:
- **Activity results are persisted**: When an activity completes, its result is saved
- **Orchestrations replay**: On restart, the orchestration replays from the beginning
- **Completed activities skip execution**: The framework uses cached results
- **Automatic resume**: The worker automatically picks up pending work on startup
### Try It Yourself
> **Tip:** To give yourself more time to stop the application during `OrderCancel`, consider increasing the loop iteration count or `Task.Delay` duration in the `OrderCancel` executor in `OrderCancelExecutors.cs`.
1. Start the application and enter an order ID (e.g., `12345`)
2. Wait for `OrderLookup` to complete, then stop the app (Ctrl+C) during `OrderCancel`
3. Restart the application
4. Observe:
- `OrderLookup` is **NOT** re-executed (result was cached)
- `OrderCancel` **restarts** (it didn't complete before the interruption)
- `SendEmail` runs after `OrderCancel` completes
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on configuring the environment, including how to install and run the Durable Task Scheduler.
## Running the Sample
```bash
cd dotnet/samples/04-hosting/DurableWorkflows/ConsoleApps/01_SequentialWorkflow
dotnet run --framework net10.0
```
### Sample Output
```text
Durable Workflow Sample
Workflow: OrderLookup -> OrderCancel -> SendEmail
Enter an order ID (or 'exit'):
> 12345
Starting workflow for order: 12345
Run ID: abc123...
[OrderLookup] Looking up order '12345'...
[OrderLookup] Found order for customer 'Jerry'
[OrderCancel] Cancelling order '12345'...
[OrderCancel] Order cancelled successfully
[SendEmail] Sending email to 'jerry@example.com'...
[SendEmail] Email sent successfully
Workflow completed!
> exit
```
@@ -0,0 +1,30 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>WorkflowConcurrency</AssemblyName>
<RootNamespace>WorkflowConcurrency</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
<PackageReference Include="Azure.AI.OpenAI" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,73 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowConcurrency;
/// <summary>
/// Parses and validates the incoming question before sending to AI agents.
/// </summary>
internal sealed class ParseQuestionExecutor() : Executor<string, string>("ParseQuestion")
{
public override ValueTask<string> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine("│ [ParseQuestion] Preparing question for AI agents...");
string formattedQuestion = message.Trim();
if (!formattedQuestion.EndsWith('?'))
{
formattedQuestion += "?";
}
Console.WriteLine($"│ [ParseQuestion] Question: \"{formattedQuestion}\"");
Console.WriteLine("│ [ParseQuestion] → Sending to Physicist and Chemist in PARALLEL...");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(formattedQuestion);
}
}
/// <summary>
/// Aggregates responses from all AI agents into a comprehensive answer.
/// This is the Fan-in point where parallel results are collected.
/// </summary>
internal sealed class AggregatorExecutor() : Executor<string[], string>("Aggregator")
{
public override ValueTask<string> HandleAsync(
string[] message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Aggregator] 📋 Received {message.Length} AI agent responses");
Console.WriteLine("│ [Aggregator] Combining into comprehensive answer...");
Console.WriteLine("│ [Aggregator] ✓ Aggregation complete!");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
string aggregatedResult = "═══════════════════════════════════════════════════════════════\n" +
" AI EXPERT PANEL RESPONSES\n" +
"═══════════════════════════════════════════════════════════════\n\n";
for (int i = 0; i < message.Length; i++)
{
string expertLabel = i == 0 ? "⚛️ PHYSICIST" : "🧪 CHEMIST";
aggregatedResult += $"{expertLabel}:\n{message[i]}\n\n";
}
aggregatedResult += "═══════════════════════════════════════════════════════════════\n" +
$"Summary: Received perspectives from {message.Length} AI experts.\n" +
"═══════════════════════════════════════════════════════════════";
return ValueTask.FromResult(aggregatedResult);
}
}
@@ -0,0 +1,114 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates the Fan-out/Fan-in pattern in a durable workflow.
// The workflow uses 4 executors: 2 class-based executors and 2 AI agents.
//
// WORKFLOW PATTERN:
//
// ParseQuestion (class-based)
// |
// +----------+----------+
// | |
// Physicist Chemist
// (AI Agent) (AI Agent)
// | |
// +----------+----------+
// |
// Aggregator (class-based)
using Azure;
using Azure.AI.OpenAI;
using Azure.Identity;
using Microsoft.Agents.AI;
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using OpenAI.Chat;
using WorkflowConcurrency;
// Configuration
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
string endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")
?? throw new InvalidOperationException("AZURE_OPENAI_ENDPOINT is not set.");
string deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT")
?? throw new InvalidOperationException("AZURE_OPENAI_DEPLOYMENT is not set.");
string? azureOpenAiKey = Environment.GetEnvironmentVariable("AZURE_OPENAI_KEY");
// Create Azure OpenAI client
AzureOpenAIClient openAiClient = !string.IsNullOrEmpty(azureOpenAiKey)
? new AzureOpenAIClient(new Uri(endpoint), new AzureKeyCredential(azureOpenAiKey))
: new AzureOpenAIClient(new Uri(endpoint), new AzureCliCredential());
ChatClient chatClient = openAiClient.GetChatClient(deploymentName);
// Define the 4 executors for the workflow
ParseQuestionExecutor parseQuestion = new();
AIAgent physicist = chatClient.AsAIAgent("You are a physics expert. Be concise (2-3 sentences).", "Physicist");
AIAgent chemist = chatClient.AsAIAgent("You are a chemistry expert. Be concise (2-3 sentences).", "Chemist");
AggregatorExecutor aggregator = new();
// Build workflow: ParseQuestion -> [Physicist, Chemist] (parallel) -> Aggregator
Workflow workflow = new WorkflowBuilder(parseQuestion)
.WithName("ExpertReview")
.AddFanOutEdge(parseQuestion, [physicist, chemist])
.AddFanInBarrierEdge([physicist, chemist], aggregator)
.Build();
// Configure and start the host
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableOptions(
options => options.Workflows.AddWorkflow(workflow),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Fan-out/Fan-in Workflow Sample");
Console.WriteLine("ParseQuestion -> [Physicist, Chemist] -> Aggregator");
Console.WriteLine();
Console.WriteLine("Enter a science question (or 'exit' to quit):");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
IWorkflowRun run = await workflowClient.RunAsync(workflow, input);
Console.WriteLine($"Run ID: {run.RunId}");
if (run is IAwaitableWorkflowRun awaitableRun)
{
string? result = await awaitableRun.WaitForCompletionAsync<string>();
Console.WriteLine("Workflow completed!");
Console.WriteLine(result);
}
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
@@ -0,0 +1,100 @@
# Concurrent Workflow Sample (Fan-Out/Fan-In)
This sample demonstrates the **fan-out/fan-in** pattern in a durable workflow, combining class-based executors with AI agents running in parallel.
## Key Concepts Demonstrated
- **Fan-out/Fan-in pattern**: Parallel execution with result aggregation
- **Mixed executor types**: Class-based executors and AI agents in the same workflow
- **AI agents as executors**: Using `ChatClient.AsAIAgent()` to create workflow-compatible agents
- **Workflow registration**: Auto-registration of agents used within workflows
- **Standalone agents**: Registering agents outside of workflows
## Overview
The sample implements an expert review workflow with four executors:
```
ParseQuestion
|
+----------+----------+
| |
Physicist Chemist
(AI Agent) (AI Agent)
| |
+----------+----------+
|
Aggregator
```
| Executor | Type | Description |
|----------|------|-------------|
| ParseQuestion | Class-based | Parses the user's question for expert review |
| Physicist | AI Agent | Provides physics perspective (runs in parallel) |
| Chemist | AI Agent | Provides chemistry perspective (runs in parallel) |
| Aggregator | Class-based | Combines expert responses into a final answer |
## Fan-Out/Fan-In Pattern
The workflow demonstrates the fan-out/fan-in pattern:
1. **Fan-out**: `ParseQuestion` sends the question to both `Physicist` and `Chemist` simultaneously
2. **Parallel execution**: Both AI agents process the question concurrently
3. **Fan-in**: `Aggregator` waits for both agents to complete, then combines their responses
This pattern is useful for:
- Gathering multiple perspectives on a problem
- Parallel processing of independent tasks
- Reducing overall execution time through concurrency
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on configuring the environment.
### Required Environment Variables
```bash
# Durable Task Scheduler (optional, defaults to localhost)
DURABLE_TASK_SCHEDULER_CONNECTION_STRING="Endpoint=http://localhost:8080;TaskHub=default;Authentication=None"
# Azure OpenAI (required)
AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/"
AZURE_OPENAI_DEPLOYMENT="gpt-4o"
AZURE_OPENAI_KEY="your-key" # Optional if using Azure CLI credentials
```
## Running the Sample
```bash
cd dotnet/samples/04-hosting/DurableWorkflows/ConsoleApps/02_ConcurrentWorkflow
dotnet run --framework net10.0
```
### Sample Output
```text
+-----------------------------------------------------------------------+
| Fan-out/Fan-in Workflow Sample (4 Executors) |
| |
| ParseQuestion -> [Physicist, Chemist] -> Aggregator |
| (class-based) (AI agents, parallel) (class-based) |
+-----------------------------------------------------------------------+
Enter a science question (or 'exit' to quit):
Question: Why is the sky blue?
Instance: abc123...
[ParseQuestion] Parsing question for expert review...
[Physicist] Analyzing from physics perspective...
[Chemist] Analyzing from chemistry perspective...
[Aggregator] Combining expert responses...
Workflow completed!
Physics perspective: The sky appears blue due to Rayleigh scattering...
Chemistry perspective: The molecular composition of our atmosphere...
Combined answer: ...
Question: exit
```
@@ -0,0 +1,29 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>ConditionalEdges</AssemblyName>
<RootNamespace>ConditionalEdges</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,85 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace ConditionalEdges;
internal sealed class Order
{
public Order(string id, decimal amount)
{
this.Id = id;
this.Amount = amount;
}
public string Id { get; }
public decimal Amount { get; }
public Customer? Customer { get; set; }
public string? PaymentReferenceNumber { get; set; }
}
public sealed record Customer(int Id, string Name, bool IsBlocked);
internal sealed class OrderIdParser() : Executor<string, Order>("OrderIdParser")
{
public override async ValueTask<Order> HandleAsync(string message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
return GetOrder(message);
}
private static Order GetOrder(string id)
{
// Simulate fetching order details
return new Order(id, 100.0m);
}
}
internal sealed class OrderEnrich() : Executor<Order, Order>("EnrichOrder")
{
public override async ValueTask<Order> HandleAsync(Order message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
message.Customer = GetCustomerForOrder(message.Id);
return message;
}
private static Customer GetCustomerForOrder(string orderId)
{
if (orderId.Contains('B'))
{
return new Customer(101, "George", true);
}
return new Customer(201, "Jerry", false);
}
}
internal sealed class PaymentProcessor() : Executor<Order, Order>("PaymentProcessor")
{
public override async ValueTask<Order> HandleAsync(Order message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
// Call payment gateway.
message.PaymentReferenceNumber = Guid.NewGuid().ToString().Substring(0, 4);
return message;
}
}
internal sealed class NotifyFraud() : Executor<Order, string>("NotifyFraud")
{
public override async ValueTask<string> HandleAsync(Order message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
// Notify fraud team.
return $"Order {message.Id} flagged as fraudulent for customer {message.Customer?.Name}.";
}
}
internal static class OrderRouteConditions
{
/// <summary>
/// Returns a condition that evaluates to true when the customer is blocked.
/// </summary>
internal static Func<Order?, bool> WhenBlocked() => order => order?.Customer?.IsBlocked == true;
/// <summary>
/// Returns a condition that evaluates to true when the customer is not blocked.
/// </summary>
internal static Func<Order?, bool> WhenNotBlocked() => order => order?.Customer?.IsBlocked == false;
}
@@ -0,0 +1,97 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates conditional edges in a workflow.
// Orders are routed to different executors based on customer status:
// - Blocked customers → NotifyFraud
// - Valid customers → PaymentProcessor
using ConditionalEdges;
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Create executor instances
OrderIdParser orderParser = new();
OrderEnrich orderEnrich = new();
PaymentProcessor paymentProcessor = new();
NotifyFraud notifyFraud = new();
// Build workflow with conditional edges
// The condition functions evaluate the Order output from OrderEnrich
WorkflowBuilder builder = new(orderParser);
builder
.AddEdge(orderParser, orderEnrich)
.AddEdge(orderEnrich, notifyFraud, condition: OrderRouteConditions.WhenBlocked())
.AddEdge(orderEnrich, paymentProcessor, condition: OrderRouteConditions.WhenNotBlocked());
Workflow auditOrder = builder.WithName("AuditOrder").Build();
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
workflowOptions => workflowOptions.AddWorkflow(auditOrder),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Enter an order ID (or 'exit'):");
Console.WriteLine("Tip: Order IDs containing 'B' are flagged as blocked customers.\n");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
await StartNewWorkflowAsync(input, auditOrder, workflowClient);
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
// Start a new workflow and wait for completion
static async Task StartNewWorkflowAsync(string orderId, Workflow workflow, IWorkflowClient client)
{
Console.WriteLine($"Starting workflow for order '{orderId}'...");
// Cast to IAwaitableWorkflowRun to access WaitForCompletionAsync
IAwaitableWorkflowRun run = (IAwaitableWorkflowRun)await client.RunAsync(workflow, orderId);
Console.WriteLine($"Run ID: {run.RunId}");
try
{
Console.WriteLine("Waiting for workflow to complete...");
string? result = await run.WaitForCompletionAsync<string>();
Console.WriteLine($"Workflow completed. {result}");
}
catch (InvalidOperationException ex)
{
Console.WriteLine($"Failed: {ex.Message}");
}
}
@@ -0,0 +1,92 @@
# Conditional Edges Workflow Sample
This sample demonstrates how to build a workflow with **conditional edges** that route execution to different paths based on runtime conditions. The workflow evaluates conditions on the output of an executor to determine which downstream executor to run.
## Key Concepts Demonstrated
- Building workflows with **conditional edges** using `AddEdge` with a `condition` parameter
- Defining reusable condition functions for routing logic
- Branching workflow execution based on data-driven decisions
- Using `ConfigureDurableWorkflows` to register workflows with dependency injection
## Overview
The sample implements an order audit workflow that routes orders differently based on whether the customer is blocked (flagged for fraud):
```
OrderIdParser --> OrderEnrich --[IsBlocked]--> NotifyFraud
|
+--[NotBlocked]--> PaymentProcessor
```
| Executor | Description |
|----------|-------------|
| OrderIdParser | Parses the order ID and retrieves order details |
| OrderEnrich | Enriches the order with customer information |
| PaymentProcessor | Processes payment for valid orders |
| NotifyFraud | Notifies the fraud team for blocked customers |
## How Conditional Edges Work
Conditional edges allow you to specify a condition function that determines whether the edge should be traversed:
```csharp
builder
.AddEdge(orderParser, orderEnrich)
.AddEdge(orderEnrich, notifyFraud, condition: OrderRouteConditions.WhenBlocked())
.AddEdge(orderEnrich, paymentProcessor, condition: OrderRouteConditions.WhenNotBlocked());
```
The condition functions receive the output of the source executor and return a boolean:
```csharp
internal static class OrderRouteConditions
{
// Routes to NotifyFraud when customer is blocked
internal static Func<Order?, bool> WhenBlocked() =>
order => order?.Customer?.IsBlocked == true;
// Routes to PaymentProcessor when customer is not blocked
internal static Func<Order?, bool> WhenNotBlocked() =>
order => order?.Customer?.IsBlocked == false;
}
```
### Routing Logic
In this sample, the routing is based on the order ID:
- Order IDs containing the letter **'B'** are associated with blocked customers → routed to `NotifyFraud`
- All other order IDs are associated with valid customers → routed to `PaymentProcessor`
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on configuring the environment, including how to install and run the Durable Task Scheduler.
## Running the Sample
```bash
cd dotnet/samples/04-hosting/DurableWorkflows/ConsoleApps/03_ConditionalEdges
dotnet run --framework net10.0
```
### Sample Output
**Valid order (routes to PaymentProcessor):**
```text
Enter an order ID (or 'exit'):
> 12345
Starting workflow for order '12345'...
Run ID: abc123...
Waiting for workflow to complete...
Workflow completed. {"Id":"12345","Amount":100.0,"Customer":{"Id":201,"Name":"Jerry","IsBlocked":false},"PaymentReferenceNumber":"a1b2"}
```
**Blocked order (routes to NotifyFraud):**
```text
Enter an order ID (or 'exit'):
> 12345B
Starting workflow for order '12345B'...
Run ID: def456...
Waiting for workflow to complete...
Workflow completed. Order 12345B flagged as fraudulent for customer George.
```
@@ -0,0 +1,30 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>WorkflowConcurrency</AssemblyName>
<RootNamespace>WorkflowConcurrency</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
<PackageReference Include="Azure.AI.OpenAI" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,73 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowConcurrency;
/// <summary>
/// Parses and validates the incoming question before sending to AI agents.
/// </summary>
internal sealed class ParseQuestionExecutor() : Executor<string, string>("ParseQuestion")
{
public override ValueTask<string> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine("│ [ParseQuestion] Preparing question for AI agents...");
string formattedQuestion = message.Trim();
if (!formattedQuestion.EndsWith('?'))
{
formattedQuestion += "?";
}
Console.WriteLine($"│ [ParseQuestion] Question: \"{formattedQuestion}\"");
Console.WriteLine("│ [ParseQuestion] → Sending to experts...");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult(formattedQuestion);
}
}
/// <summary>
/// Aggregates responses from multiple AI agents into a unified response.
/// This executor collects all expert opinions and synthesizes them.
/// </summary>
internal sealed class ResponseAggregatorExecutor() : Executor<string[], string>("ResponseAggregator")
{
public override ValueTask<string> HandleAsync(
string[] message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [Aggregator] 📋 Received {message.Length} AI agent responses");
Console.WriteLine("│ [Aggregator] Combining into comprehensive answer...");
Console.WriteLine("│ [Aggregator] ✓ Aggregation complete!");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
string aggregatedResult = "═══════════════════════════════════════════════════════════════\n" +
" AI EXPERT PANEL RESPONSES\n" +
"═══════════════════════════════════════════════════════════════\n\n";
for (int i = 0; i < message.Length; i++)
{
string expertLabel = i == 0 ? "⚛️ PHYSICIST" : "🧪 CHEMIST";
aggregatedResult += $"{expertLabel}:\n{message[i]}\n\n";
}
aggregatedResult += "═══════════════════════════════════════════════════════════════\n" +
$"Summary: Received perspectives from {message.Length} AI experts.\n" +
"═══════════════════════════════════════════════════════════════";
return ValueTask.FromResult(aggregatedResult);
}
}
@@ -0,0 +1,133 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates the THREE ways to configure durable agents and workflows:
//
// 1. ConfigureDurableAgents() - For standalone agents only
// 2. ConfigureDurableWorkflows() - For workflows only
// 3. ConfigureDurableOptions() - For both agents AND workflows
//
// KEY: All methods can be called MULTIPLE times - configurations are ADDITIVE.
using Azure;
using Azure.AI.OpenAI;
using Azure.Identity;
using Microsoft.Agents.AI;
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using OpenAI.Chat;
using WorkflowConcurrency;
// Configuration
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
string endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")
?? throw new InvalidOperationException("AZURE_OPENAI_ENDPOINT is not set.");
string deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT")
?? throw new InvalidOperationException("AZURE_OPENAI_DEPLOYMENT is not set.");
string? azureOpenAiKey = Environment.GetEnvironmentVariable("AZURE_OPENAI_KEY");
// Create AI agents
AzureOpenAIClient openAiClient = !string.IsNullOrEmpty(azureOpenAiKey)
? new AzureOpenAIClient(new Uri(endpoint), new AzureKeyCredential(azureOpenAiKey))
: new AzureOpenAIClient(new Uri(endpoint), new AzureCliCredential());
ChatClient chatClient = openAiClient.GetChatClient(deploymentName);
AIAgent biologist = chatClient.AsAIAgent("You are a biology expert. Explain concepts clearly in 2-3 sentences.", "Biologist");
AIAgent physicist = chatClient.AsAIAgent("You are a physics expert. Explain concepts clearly in 2-3 sentences.", "Physicist");
AIAgent chemist = chatClient.AsAIAgent("You are a chemistry expert. Explain concepts clearly in 2-3 sentences.", "Chemist");
// Create workflows
ParseQuestionExecutor questionParser = new();
ResponseAggregatorExecutor responseAggregator = new();
Workflow physicsWorkflow = new WorkflowBuilder(questionParser)
.WithName("PhysicsExpertReview")
.AddEdge(questionParser, physicist)
.Build();
Workflow expertTeamWorkflow = new WorkflowBuilder(questionParser)
.WithName("ExpertTeamReview")
.AddFanOutEdge(questionParser, [biologist, physicist])
.AddFanInBarrierEdge([biologist, physicist], responseAggregator)
.Build();
Workflow chemistryWorkflow = new WorkflowBuilder(questionParser)
.WithName("ChemistryExpertReview")
.AddEdge(questionParser, chemist)
.Build();
// Configure services - demonstrating all 3 methods (each can be called multiple times)
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
// METHOD 1: ConfigureDurableAgents - for standalone agents only
services.ConfigureDurableAgents(
options => options.AddAIAgent(biologist),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
// METHOD 2: ConfigureDurableWorkflows - for workflows only
services.ConfigureDurableWorkflows(options => options.AddWorkflow(physicsWorkflow));
// METHOD 3: ConfigureDurableOptions - for both agents AND workflows
services.ConfigureDurableOptions(options =>
{
options.Agents.AddAIAgent(chemist);
options.Workflows.AddWorkflow(expertTeamWorkflow);
});
// Second call to ConfigureDurableOptions (additive - adds to existing config)
services.ConfigureDurableOptions(options => options.Workflows.AddWorkflow(chemistryWorkflow));
})
.Build();
await host.StartAsync();
IServiceProvider services = host.Services;
IWorkflowClient workflowClient = services.GetRequiredService<IWorkflowClient>();
// DEMO 1: Direct agent conversation (standalone agents)
Console.WriteLine("\n═══ DEMO 1: Direct Agent Conversation ═══\n");
AIAgent biologistProxy = services.GetRequiredKeyedService<AIAgent>("Biologist");
AgentSession session = await biologistProxy.CreateSessionAsync();
AgentResponse response = await biologistProxy.RunAsync("What is photosynthesis?", session);
Console.WriteLine($"🧬 Biologist: {response.Text}\n");
AIAgent chemistProxy = services.GetRequiredKeyedService<AIAgent>("Chemist");
session = await chemistProxy.CreateSessionAsync();
response = await chemistProxy.RunAsync("What is a chemical bond?", session);
Console.WriteLine($"🧪 Chemist: {response.Text}\n");
// DEMO 2: Single-agent workflow
Console.WriteLine("═══ DEMO 2: Single-Agent Workflow ═══\n");
await RunWorkflowAsync(workflowClient, physicsWorkflow, "What is the relationship between energy and mass?");
// DEMO 3: Multi-agent workflow
Console.WriteLine("═══ DEMO 3: Multi-Agent Workflow ═══\n");
await RunWorkflowAsync(workflowClient, expertTeamWorkflow, "How does radiation affect living cells?");
// DEMO 4: Workflow from second ConfigureDurableOptions call
Console.WriteLine("═══ DEMO 4: Workflow (added via 2nd ConfigureDurableOptions) ═══\n");
await RunWorkflowAsync(workflowClient, chemistryWorkflow, "What happens during combustion?");
Console.WriteLine("\n✅ All demos completed!");
await host.StopAsync();
// Helper method
static async Task RunWorkflowAsync(IWorkflowClient client, Workflow workflow, string question)
{
Console.WriteLine($"📋 {workflow.Name}: \"{question}\"");
IWorkflowRun run = await client.RunAsync(workflow, question);
if (run is IAwaitableWorkflowRun awaitable)
{
string? result = await awaitable.WaitForCompletionAsync<string>();
Console.WriteLine($"✅ {result}\n");
}
}
@@ -0,0 +1,28 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>WorkflowEvents</AssemblyName>
<RootNamespace>WorkflowEvents</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,129 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowEvents;
// ═══════════════════════════════════════════════════════════════════════════════
// Custom event types - callers observe these via WatchStreamAsync
// ═══════════════════════════════════════════════════════════════════════════════
internal sealed class OrderLookupStartedEvent(string orderId) : WorkflowEvent(orderId)
{
public string OrderId { get; } = orderId;
}
internal sealed class OrderFoundEvent(string customerName) : WorkflowEvent(customerName)
{
public string CustomerName { get; } = customerName;
}
internal sealed class CancellationProgressEvent(int percentComplete, string status) : WorkflowEvent(status)
{
public int PercentComplete { get; } = percentComplete;
public string Status { get; } = status;
}
internal sealed class OrderCancelledEvent() : WorkflowEvent("Order cancelled");
internal sealed class EmailSentEvent(string email) : WorkflowEvent(email)
{
public string Email { get; } = email;
}
// ═══════════════════════════════════════════════════════════════════════════════
// Domain models
// ═══════════════════════════════════════════════════════════════════════════════
internal sealed record Order(string Id, DateTime OrderDate, bool IsCancelled, string? CancelReason, Customer Customer);
internal sealed record Customer(string Name, string Email);
// ═══════════════════════════════════════════════════════════════════════════════
// Executors - emit events via AddEventAsync and YieldOutputAsync
// ═══════════════════════════════════════════════════════════════════════════════
/// <summary>
/// Looks up an order by ID, emitting progress events.
/// </summary>
internal sealed class OrderLookup() : Executor<string, Order>("OrderLookup")
{
public override async ValueTask<Order> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await context.AddEventAsync(new OrderLookupStartedEvent(message), cancellationToken);
// Simulate database lookup
await Task.Delay(TimeSpan.FromSeconds(1), cancellationToken);
Order order = new(
Id: message,
OrderDate: DateTime.UtcNow.AddDays(-1),
IsCancelled: false,
CancelReason: "Customer requested cancellation",
Customer: new Customer(Name: "Jerry", Email: "jerry@example.com"));
await context.AddEventAsync(new OrderFoundEvent(order.Customer.Name), cancellationToken);
// YieldOutputAsync emits a WorkflowOutputEvent observable via streaming
await context.YieldOutputAsync(order, cancellationToken);
return order;
}
}
/// <summary>
/// Cancels an order, emitting progress events during the multi-step process.
/// </summary>
internal sealed class OrderCancel() : Executor<Order, Order>("OrderCancel")
{
public override async ValueTask<Order> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await context.AddEventAsync(new CancellationProgressEvent(0, "Starting cancellation"), cancellationToken);
// Simulate a multi-step cancellation process
await Task.Delay(TimeSpan.FromMilliseconds(500), cancellationToken);
await context.AddEventAsync(new CancellationProgressEvent(33, "Contacting payment provider"), cancellationToken);
await Task.Delay(TimeSpan.FromMilliseconds(500), cancellationToken);
await context.AddEventAsync(new CancellationProgressEvent(66, "Processing refund"), cancellationToken);
await Task.Delay(TimeSpan.FromMilliseconds(500), cancellationToken);
Order cancelledOrder = message with { IsCancelled = true };
await context.AddEventAsync(new CancellationProgressEvent(100, "Complete"), cancellationToken);
await context.AddEventAsync(new OrderCancelledEvent(), cancellationToken);
await context.YieldOutputAsync(cancelledOrder, cancellationToken);
return cancelledOrder;
}
}
/// <summary>
/// Sends a cancellation confirmation email, emitting an event on completion.
/// </summary>
internal sealed class SendEmail() : Executor<Order, string>("SendEmail")
{
public override async ValueTask<string> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// Simulate sending email
await Task.Delay(TimeSpan.FromMilliseconds(500), cancellationToken);
string result = $"Cancellation email sent for order {message.Id} to {message.Customer.Email}.";
await context.AddEventAsync(new EmailSentEvent(message.Customer.Email), cancellationToken);
await context.YieldOutputAsync(result, cancellationToken);
return result;
}
}
@@ -0,0 +1,138 @@
// Copyright (c) Microsoft. All rights reserved.
// ═══════════════════════════════════════════════════════════════════════════════
// SAMPLE: Workflow Events and Streaming
// ═══════════════════════════════════════════════════════════════════════════════
//
// This sample demonstrates how to use IWorkflowContext event methods in executors
// and stream events from the caller side:
//
// 1. AddEventAsync - Emit custom events that callers can observe in real-time
// 2. StreamAsync - Start a workflow and obtain a streaming handle
// 3. WatchStreamAsync - Observe events as they occur (custom, framework, and terminal)
//
// The sample uses IWorkflowClient.StreamAsync to start a workflow and
// WatchStreamAsync to observe events as they occur in real-time.
//
// Workflow: OrderLookup -> OrderCancel -> SendEmail
// ═══════════════════════════════════════════════════════════════════════════════
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using WorkflowEvents;
// Get DTS connection string from environment variable
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Define executors and build workflow
OrderLookup orderLookup = new();
OrderCancel orderCancel = new();
SendEmail sendEmail = new();
Workflow cancelOrder = new WorkflowBuilder(orderLookup)
.WithName("CancelOrder")
.WithDescription("Cancel an order and notify the customer")
.AddEdge(orderLookup, orderCancel)
.AddEdge(orderCancel, sendEmail)
.Build();
// Configure host with durable workflow support
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
workflowOptions => workflowOptions.AddWorkflow(cancelOrder),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Workflow Events Demo - Enter order ID (or 'exit'):");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
await RunWorkflowWithStreamingAsync(input, cancelOrder, workflowClient);
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
// Runs a workflow and streams events as they occur
static async Task RunWorkflowWithStreamingAsync(string orderId, Workflow workflow, IWorkflowClient client)
{
// StreamAsync starts the workflow and returns a streaming handle for observing events
IStreamingWorkflowRun run = await client.StreamAsync(workflow, orderId);
Console.WriteLine($"Started run: {run.RunId}");
// WatchStreamAsync yields events as they're emitted by executors
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
Console.WriteLine($" New event received at {DateTime.Now:HH:mm:ss.ffff} ({evt.GetType().Name})");
switch (evt)
{
// Custom domain events (emitted via AddEventAsync)
case OrderLookupStartedEvent e:
WriteColored($" [Lookup] Looking up order {e.OrderId}", ConsoleColor.Cyan);
break;
case OrderFoundEvent e:
WriteColored($" [Lookup] Found: {e.CustomerName}", ConsoleColor.Cyan);
break;
case CancellationProgressEvent e:
WriteColored($" [Cancel] {e.PercentComplete}% - {e.Status}", ConsoleColor.Yellow);
break;
case OrderCancelledEvent:
WriteColored(" [Cancel] Done", ConsoleColor.Yellow);
break;
case EmailSentEvent e:
WriteColored($" [Email] Sent to {e.Email}", ConsoleColor.Magenta);
break;
case WorkflowOutputEvent e:
WriteColored($" [Output] {e.ExecutorId}", ConsoleColor.DarkGray);
break;
// Workflow completion
case DurableWorkflowCompletedEvent e:
WriteColored($" Completed: {e.Result}", ConsoleColor.Green);
break;
case DurableWorkflowFailedEvent e:
WriteColored($" Failed: {e.ErrorMessage}", ConsoleColor.Red);
break;
}
}
}
static void WriteColored(string message, ConsoleColor color)
{
Console.ForegroundColor = color;
Console.WriteLine(message);
Console.ResetColor();
}
@@ -0,0 +1,127 @@
# Workflow Events Sample
This sample demonstrates how to use workflow events and streaming in durable workflows.
## What it demonstrates
1. **Custom Events** (`AddEventAsync`) — Executors emit domain-specific events during execution
2. **Event Streaming** (`StreamAsync` / `WatchStreamAsync`) — Callers observe events in real-time as the workflow progresses
3. **Framework Events** — Automatic `ExecutorInvokedEvent`, `ExecutorCompletedEvent`, and `WorkflowOutputEvent` events emitted by the framework
## Emitting Custom Events
Executors can emit custom domain events during execution using the `IWorkflowContext` instance passed to `HandleAsync`. These events are streamed to callers in real-time via `WatchStreamAsync`.
### Defining a custom event
Create a class that inherits from `WorkflowEvent`. Pass any data payload to the base constructor:
```csharp
public class CancellationProgressEvent(int percentComplete, string status) : WorkflowEvent(status)
{
public int PercentComplete { get; } = percentComplete;
public string Status { get; } = status;
}
```
### Emitting the event from an executor
Call `AddEventAsync` on the `IWorkflowContext` inside your executor's `HandleAsync` method:
```csharp
public override async ValueTask<Order> HandleAsync(
Order message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await context.AddEventAsync(new CancellationProgressEvent(33, "Processing refund"), cancellationToken);
// ... rest of the executor logic
}
```
### Observing events from the caller
Use `StreamAsync` to start the workflow and `WatchStreamAsync` to observe events. Pattern match on your custom event types:
```csharp
IStreamingWorkflowRun run = await workflowClient.StreamAsync(workflow, input);
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
switch (evt)
{
case CancellationProgressEvent e:
Console.WriteLine($"{e.PercentComplete}% - {e.Status}");
break;
}
}
```
## Workflow Structure
```
OrderLookup → OrderCancel → SendEmail
```
Each executor emits custom events during execution:
- `OrderLookup` emits `OrderLookupStartedEvent` and `OrderFoundEvent`
- `OrderCancel` emits `CancellationProgressEvent` (with percentage) and `OrderCancelledEvent`
- `SendEmail` emits `EmailSentEvent`
## Prerequisites
- [Durable Task Scheduler](https://learn.microsoft.com/en-us/azure/azure-functions/durable/durable-task-scheduler/durable-task-scheduler) running locally or in Azure
- Set the `DURABLE_TASK_SCHEDULER_CONNECTION_STRING` environment variable (defaults to local emulator)
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for more information on how to configure the environment, including how to install and run common sample dependencies.
## Running the sample
```bash
dotnet run
```
Enter an order ID at the prompt to start a workflow and watch events stream in real-time:
```text
> order-42
Started run: b6ba4d19...
New event received at 13:27:41.4956 (ExecutorInvokedEvent)
New event received at 13:27:41.5019 (OrderLookupStartedEvent)
[Lookup] Looking up order order-42
New event received at 13:27:41.5025 (OrderFoundEvent)
[Lookup] Found: Jerry
New event received at 13:27:41.5026 (ExecutorCompletedEvent)
New event received at 13:27:41.5026 (WorkflowOutputEvent)
[Output] OrderLookup
New event received at 13:27:43.0772 (ExecutorInvokedEvent)
New event received at 13:27:43.0773 (CancellationProgressEvent)
[Cancel] 0% - Starting cancellation
New event received at 13:27:43.0775 (CancellationProgressEvent)
[Cancel] 33% - Contacting payment provider
New event received at 13:27:43.0776 (CancellationProgressEvent)
[Cancel] 66% - Processing refund
New event received at 13:27:43.0777 (CancellationProgressEvent)
[Cancel] 100% - Complete
New event received at 13:27:43.0779 (OrderCancelledEvent)
[Cancel] Done
New event received at 13:27:43.0780 (ExecutorCompletedEvent)
New event received at 13:27:43.0780 (WorkflowOutputEvent)
[Output] OrderCancel
New event received at 13:27:43.6610 (ExecutorInvokedEvent)
New event received at 13:27:43.6611 (EmailSentEvent)
[Email] Sent to jerry@example.com
New event received at 13:27:43.6613 (ExecutorCompletedEvent)
New event received at 13:27:43.6613 (WorkflowOutputEvent)
[Output] SendEmail
New event received at 13:27:43.6619 (DurableWorkflowCompletedEvent)
Completed: Cancellation email sent for order order-42 to jerry@example.com.
```
### Viewing Workflows in the DTS Dashboard
After running a workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to inspect the workflow execution and events.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
@@ -0,0 +1,29 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>WorkflowSharedState</AssemblyName>
<RootNamespace>WorkflowSharedState</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,184 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowSharedState;
// ═══════════════════════════════════════════════════════════════════════════════
// Domain models
// ═══════════════════════════════════════════════════════════════════════════════
/// <summary>
/// The primary order data passed through the pipeline via return values.
/// </summary>
internal sealed record OrderDetails(string OrderId, string CustomerName, decimal Amount, DateTime OrderDate);
/// <summary>
/// Cross-cutting audit trail accumulated in shared state across executors.
/// Each executor appends its step name and timestamp. This data does not flow
/// through return values — it lives only in shared state.
/// </summary>
internal sealed record AuditEntry(string Step, string Timestamp, string Detail);
// ═══════════════════════════════════════════════════════════════════════════════
// Executors
// ═══════════════════════════════════════════════════════════════════════════════
/// <summary>
/// Validates the order and writes the initial audit entry and tax rate to shared state.
/// The order details are returned as the executor output (normal message flow),
/// while the audit trail and tax rate are stored in shared state (side-channel).
/// If the order ID starts with "INVALID", the executor halts the workflow early
/// using <see cref="IWorkflowContext.RequestHaltAsync"/>.
/// </summary>
[YieldsOutput(typeof(string))]
internal sealed class ValidateOrder() : Executor<string, OrderDetails>("ValidateOrder")
{
public override async ValueTask<OrderDetails> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await Task.Delay(TimeSpan.FromMilliseconds(200), cancellationToken);
// Halt the workflow early if the order ID is invalid.
// No downstream executors will run after this.
if (message.StartsWith("INVALID", StringComparison.OrdinalIgnoreCase))
{
await context.YieldOutputAsync($"Order '{message}' failed validation. Halting workflow.", cancellationToken);
await context.RequestHaltAsync();
return new OrderDetails(message, "Unknown", 0, DateTime.UtcNow);
}
OrderDetails details = new(message, "Jerry", 249.99m, DateTime.UtcNow);
// Store the tax rate in shared state — downstream ProcessPayment reads it
// without needing it in the message chain.
await context.QueueStateUpdateAsync("taxRate", 0.085m, cancellationToken: cancellationToken);
Console.WriteLine(" Wrote to shared state: taxRate = 8.5%");
// Start the audit trail in shared state
AuditEntry audit = new("ValidateOrder", DateTime.UtcNow.ToString("o"), $"Validated order {message}");
await context.QueueStateUpdateAsync("auditValidate", audit, cancellationToken: cancellationToken);
Console.WriteLine(" Wrote to shared state: auditValidate");
await context.YieldOutputAsync($"Order '{message}' validated. Customer: {details.CustomerName}, Amount: {details.Amount:C}", cancellationToken);
return details;
}
}
/// <summary>
/// Enriches the order with shipping information.
/// Reads the audit trail from shared state and appends its own entry.
/// Uses ReadOrInitStateAsync to lazily initialize a shipping tier.
/// Demonstrates custom scopes by writing shipping details under the "shipping" scope.
/// </summary>
[YieldsOutput(typeof(string))]
internal sealed class EnrichOrder() : Executor<OrderDetails, OrderDetails>("EnrichOrder")
{
public override async ValueTask<OrderDetails> HandleAsync(
OrderDetails message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await Task.Delay(TimeSpan.FromMilliseconds(200), cancellationToken);
// Use ReadOrInitStateAsync — only initializes if no value exists yet
string shippingTier = await context.ReadOrInitStateAsync(
"shippingTier",
() => "Express",
cancellationToken: cancellationToken);
Console.WriteLine($" Read from shared state: shippingTier = {shippingTier}");
// Write carrier under a custom "shipping" scope.
// This keeps the key separate from keys written without a scope,
// so "carrier" here won't collide with a "carrier" key written elsewhere.
await context.QueueStateUpdateAsync("carrier", "Contoso Express", scopeName: "shipping", cancellationToken: cancellationToken);
Console.WriteLine(" Wrote to shared state: carrier = Contoso Express (scope: shipping)");
// Verify we can read the audit entry from the previous step
AuditEntry? previousAudit = await context.ReadStateAsync<AuditEntry>("auditValidate", cancellationToken: cancellationToken);
string auditStatus = previousAudit is not null ? $"(previous step: {previousAudit.Step})" : "(no prior audit)";
Console.WriteLine($" Read from shared state: auditValidate {auditStatus}");
// Append our own audit entry
AuditEntry audit = new("EnrichOrder", DateTime.UtcNow.ToString("o"), $"Enriched with {shippingTier} shipping {auditStatus}");
await context.QueueStateUpdateAsync("auditEnrich", audit, cancellationToken: cancellationToken);
Console.WriteLine(" Wrote to shared state: auditEnrich");
await context.YieldOutputAsync($"Order enriched. Shipping: {shippingTier} {auditStatus}", cancellationToken);
return message;
}
}
/// <summary>
/// Processes payment using the tax rate from shared state (written by ValidateOrder).
/// The tax rate is side-channel data — it doesn't flow through return values.
/// </summary>
internal sealed class ProcessPayment() : Executor<OrderDetails, string>("ProcessPayment")
{
public override async ValueTask<string> HandleAsync(
OrderDetails message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await Task.Delay(TimeSpan.FromMilliseconds(300), cancellationToken);
// Read tax rate written by ValidateOrder — not available in the message chain
decimal taxRate = await context.ReadOrInitStateAsync("taxRate", () => 0.0m, cancellationToken: cancellationToken);
Console.WriteLine($" Read from shared state: taxRate = {taxRate:P1}");
decimal tax = message.Amount * taxRate;
decimal total = message.Amount + tax;
string paymentRef = $"PAY-{Guid.NewGuid():N}"[..16];
// Append audit entry
AuditEntry audit = new("ProcessPayment", DateTime.UtcNow.ToString("o"), $"Charged {total:C} (tax: {tax:C})");
await context.QueueStateUpdateAsync("auditPayment", audit, cancellationToken: cancellationToken);
Console.WriteLine(" Wrote to shared state: auditPayment");
await context.YieldOutputAsync($"Payment processed. Total: {total:C} (tax: {tax:C}). Ref: {paymentRef}", cancellationToken);
return paymentRef;
}
}
/// <summary>
/// Generates the final invoice by reading the full audit trail from shared state.
/// Demonstrates reading multiple state entries written by different executors
/// and clearing a scope with <see cref="IWorkflowContext.QueueClearScopeAsync(string?, CancellationToken)"/>.
/// </summary>
internal sealed class GenerateInvoice() : Executor<string, string>("GenerateInvoice")
{
public override async ValueTask<string> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
// Read the full audit trail from shared state — each step wrote its own entry
AuditEntry? validateAudit = await context.ReadStateAsync<AuditEntry>("auditValidate", cancellationToken: cancellationToken);
AuditEntry? enrichAudit = await context.ReadStateAsync<AuditEntry>("auditEnrich", cancellationToken: cancellationToken);
AuditEntry? paymentAudit = await context.ReadStateAsync<AuditEntry>("auditPayment", cancellationToken: cancellationToken);
int auditCount = new[] { validateAudit, enrichAudit, paymentAudit }.Count(a => a is not null);
Console.WriteLine($" Read from shared state: {auditCount} audit entries");
// Read carrier from the "shipping" scope (written by EnrichOrder)
string? carrier = await context.ReadStateAsync<string>("carrier", scopeName: "shipping", cancellationToken: cancellationToken);
Console.WriteLine($" Read from shared state: carrier = {carrier} (scope: shipping)");
// Clear the "shipping" scope — no longer needed after invoice generation.
await context.QueueClearScopeAsync("shipping", cancellationToken);
Console.WriteLine(" Cleared shared state scope: shipping");
string auditSummary = string.Join(" → ", new[]
{
validateAudit?.Step, enrichAudit?.Step, paymentAudit?.Step
}.Where(s => s is not null));
return $"Invoice complete. Payment: {message}. Audit trail: [{auditSummary}]";
}
}
@@ -0,0 +1,117 @@
// Copyright (c) Microsoft. All rights reserved.
// ═══════════════════════════════════════════════════════════════════════════════
// SAMPLE: Shared State During Workflow Execution
// ═══════════════════════════════════════════════════════════════════════════════
//
// This sample demonstrates how executors in a durable workflow can share state
// via IWorkflowContext. State is persisted across supersteps and survives
// process restarts because the orchestration passes it to each activity.
//
// Key concepts:
// 1. QueueStateUpdateAsync - Write a value to shared state
// 2. ReadStateAsync - Read a value written by a previous executor
// 3. ReadOrInitStateAsync - Read or lazily initialize a state value
// 4. QueueClearScopeAsync - Clear all entries under a scope
// 5. RequestHaltAsync - Stop the workflow early (e.g., validation failure)
//
// Workflow: ValidateOrder -> EnrichOrder -> ProcessPayment -> GenerateInvoice
//
// Return values carry primary business data through the pipeline (OrderDetails,
// payment ref). Shared state carries side-channel data that doesn't belong in
// the message chain: a tax rate (set by ValidateOrder, read by ProcessPayment)
// and an audit trail (each executor appends its own entry).
// ═══════════════════════════════════════════════════════════════════════════════
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using WorkflowSharedState;
// Get DTS connection string from environment variable
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Define executors
ValidateOrder validateOrder = new();
EnrichOrder enrichOrder = new();
ProcessPayment processPayment = new();
GenerateInvoice generateInvoice = new();
// Build the workflow: ValidateOrder -> EnrichOrder -> ProcessPayment -> GenerateInvoice
Workflow orderPipeline = new WorkflowBuilder(validateOrder)
.WithName("OrderPipeline")
.WithDescription("Order processing pipeline with shared state across executors")
.AddEdge(validateOrder, enrichOrder)
.AddEdge(enrichOrder, processPayment)
.AddEdge(processPayment, generateInvoice)
.Build();
// Configure host with durable workflow support
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
workflowOptions => workflowOptions.AddWorkflow(orderPipeline),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Shared State Workflow Demo");
Console.WriteLine("Workflow: ValidateOrder -> EnrichOrder -> ProcessPayment -> GenerateInvoice");
Console.WriteLine();
Console.WriteLine("Enter an order ID (or 'exit'):");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
// Start the workflow and stream events to see shared state in action
IStreamingWorkflowRun run = await workflowClient.StreamAsync(orderPipeline, input);
Console.WriteLine($"Started run: {run.RunId}");
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
switch (evt)
{
case WorkflowOutputEvent e:
Console.WriteLine($" [Output] {e.ExecutorId}: {e.Data}");
break;
case DurableWorkflowCompletedEvent e:
Console.WriteLine($" Completed: {e.Result}");
break;
case DurableWorkflowFailedEvent e:
Console.WriteLine($" Failed: {e.ErrorMessage}");
break;
}
}
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
@@ -0,0 +1,71 @@
# Shared State Workflow Sample
This sample demonstrates how executors in a durable workflow can share state via `IWorkflowContext`. State written by one executor is accessible to all downstream executors, persisted across supersteps, and survives process restarts.
## Key Concepts Demonstrated
- Writing state with `QueueStateUpdateAsync` — executors store data for downstream executors
- Reading state with `ReadStateAsync` — executors access data written by earlier executors
- Lazy initialization with `ReadOrInitStateAsync` — initialize state only if not already present
- Custom scopes with `scopeName` — partition state into isolated namespaces (e.g., `"shipping"`)
- Clearing scopes with `QueueClearScopeAsync` — remove all entries under a scope when no longer needed
- Early termination with `RequestHaltAsync` — halt the workflow when validation fails
- State persistence across supersteps — the orchestration passes shared state to each executor
- Event streaming with `IStreamingWorkflowRun` — observe executor progress in real time
## Workflow
**OrderPipeline**: `ValidateOrder``EnrichOrder``ProcessPayment``GenerateInvoice`
Return values carry primary business data through the pipeline (`OrderDetails``OrderDetails` → payment ref → invoice string). Shared state carries side-channel data that doesn't belong in the message chain:
| Executor | Returns (message flow) | Reads from State | Writes to State |
|----------|----------------------|-----------------|-----------------|
| **ValidateOrder** | `OrderDetails` | — | `taxRate`, `auditValidate` |
| **EnrichOrder** | `OrderDetails` (pass-through) | `auditValidate` | `shippingTier`, `auditEnrich`, `carrier` (scope: shipping) |
| **ProcessPayment** | payment ref string | `taxRate` | `auditPayment` |
| **GenerateInvoice** | invoice string | `auditValidate`, `auditEnrich`, `auditPayment`, `carrier` (scope: shipping) | clears `shipping` scope |
> [!NOTE]
> `EnrichOrder` writes `carrier` under the `"shipping"` scope using `scopeName: "shipping"`. This keeps the key separate from keys written without a scope, so `"carrier"` in the `"shipping"` scope won't collide with a `"carrier"` key written elsewhere.
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for more information on how to configure the environment, including how to install and run common sample dependencies.
## Running the Sample
```bash
dotnet run
```
Enter an order ID when prompted. The workflow will process the order through all four executors, streaming events as they occur:
```text
> ORD-001
Started run: abc123
Wrote to shared state: taxRate = 8.5%
Wrote to shared state: auditValidate
[Output] ValidateOrder: Order 'ORD-001' validated. Customer: Jerry, Amount: $249.99
Read from shared state: shippingTier = Express
Wrote to shared state: carrier = Contoso Express (scope: shipping)
Read from shared state: auditValidate (previous step: ValidateOrder)
Wrote to shared state: auditEnrich
[Output] EnrichOrder: Order enriched. Shipping: Express (previous step: ValidateOrder)
Read from shared state: taxRate = 8.5%
Wrote to shared state: auditPayment
[Output] ProcessPayment: Payment processed. Total: $271.24 (tax: $21.25). Ref: PAY-abc123def456
Read from shared state: 3 audit entries
Read from shared state: carrier = Contoso Express (scope: shipping)
Cleared shared state scope: shipping
[Output] GenerateInvoice: Invoice complete. Payment: "PAY-abc123def456". Audit trail: [ValidateOrder → EnrichOrder → ProcessPayment]
Completed: Invoice complete. Payment: "PAY-abc123def456". Audit trail: [ValidateOrder → EnrichOrder → ProcessPayment]
```
### Viewing Workflows in the DTS Dashboard
After running a workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to inspect the orchestration status, executor inputs/outputs, and events.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
To inspect shared state in the dashboard, click on an executor to view its input and output. The input contains a snapshot of the shared state the executor ran with, and the output includes any state updates it made (as `stateUpdates` with scoped keys).
@@ -0,0 +1,28 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>SubWorkflows</AssemblyName>
<RootNamespace>SubWorkflows</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,232 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace SubWorkflows;
/// <summary>
/// Event emitted when the fraud check risk score is calculated.
/// </summary>
internal sealed class FraudRiskAssessedEvent(int riskScore) : WorkflowEvent($"Risk score: {riskScore}/100")
{
public int RiskScore => riskScore;
}
/// <summary>
/// Represents an order being processed through the workflow.
/// </summary>
internal sealed class OrderInfo
{
public required string OrderId { get; set; }
public decimal Amount { get; set; }
public string? PaymentTransactionId { get; set; }
public string? TrackingNumber { get; set; }
public string? Carrier { get; set; }
}
// Main workflow executors
/// <summary>
/// Entry point executor that receives the order ID and creates an OrderInfo object.
/// </summary>
internal sealed class OrderReceived() : Executor<string, OrderInfo>("OrderReceived")
{
public override ValueTask<OrderInfo> HandleAsync(string message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine($"[OrderReceived] Processing order '{message}'");
Console.ResetColor();
OrderInfo order = new()
{
OrderId = message,
Amount = 99.99m // Simulated order amount
};
return ValueTask.FromResult(order);
}
}
/// <summary>
/// Final executor that outputs the completed order summary.
/// </summary>
internal sealed class OrderCompleted() : Executor<OrderInfo, string>("OrderCompleted")
{
public override ValueTask<string> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine("┌─────────────────────────────────────────────────────────────────┐");
Console.WriteLine($"│ [OrderCompleted] Order '{message.OrderId}' successfully processed!");
Console.WriteLine($"│ Payment: {message.PaymentTransactionId}");
Console.WriteLine($"│ Shipping: {message.Carrier} - {message.TrackingNumber}");
Console.WriteLine("└─────────────────────────────────────────────────────────────────┘");
Console.ResetColor();
return ValueTask.FromResult($"Order {message.OrderId} completed. Tracking: {message.TrackingNumber}");
}
}
// Payment sub-workflow executors
/// <summary>
/// Validates payment information for an order.
/// </summary>
internal sealed class ValidatePayment() : Executor<OrderInfo, OrderInfo>("ValidatePayment")
{
public override async ValueTask<OrderInfo> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($" [Payment/ValidatePayment] Validating payment for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($" [Payment/ValidatePayment] Payment validated for ${message.Amount}");
Console.ResetColor();
return message;
}
}
/// <summary>
/// Charges the payment for an order.
/// </summary>
internal sealed class ChargePayment() : Executor<OrderInfo, OrderInfo>("ChargePayment")
{
public override async ValueTask<OrderInfo> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($" [Payment/ChargePayment] Charging ${message.Amount} for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
message.PaymentTransactionId = $"TXN-{Guid.NewGuid().ToString("N")[..8].ToUpperInvariant()}";
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($" [Payment/ChargePayment] ✓ Payment processed: {message.PaymentTransactionId}");
Console.ResetColor();
return message;
}
}
// FraudCheck sub-sub-workflow executors (nested inside Payment)
/// <summary>
/// Analyzes transaction patterns for potential fraud.
/// </summary>
internal sealed class AnalyzePatterns() : Executor<OrderInfo, OrderInfo>("AnalyzePatterns")
{
public override async ValueTask<OrderInfo> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($" [Payment/FraudCheck/AnalyzePatterns] Analyzing patterns for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
// Store analysis results in shared state for the next executor in this sub-workflow
int patternsFound = new Random().Next(0, 5);
await context.QueueStateUpdateAsync("patternsFound", patternsFound, cancellationToken: cancellationToken);
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($" [Payment/FraudCheck/AnalyzePatterns] ✓ Pattern analysis complete ({patternsFound} suspicious patterns)");
Console.ResetColor();
return message;
}
}
/// <summary>
/// Calculates a risk score for the transaction.
/// </summary>
internal sealed class CalculateRiskScore() : Executor<OrderInfo, OrderInfo>("CalculateRiskScore")
{
public override async ValueTask<OrderInfo> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($" [Payment/FraudCheck/CalculateRiskScore] Calculating risk score for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
// Read the pattern count from shared state (written by AnalyzePatterns)
int patternsFound = await context.ReadStateAsync<int>("patternsFound", cancellationToken: cancellationToken);
int riskScore = Math.Min(patternsFound * 20 + new Random().Next(1, 20), 100);
// Emit a workflow event from within a nested sub-workflow
await context.AddEventAsync(new FraudRiskAssessedEvent(riskScore), cancellationToken);
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($" [Payment/FraudCheck/CalculateRiskScore] ✓ Risk score: {riskScore}/100 (based on {patternsFound} patterns)");
Console.ResetColor();
return message;
}
}
// Shipping sub-workflow executors
/// <summary>
/// Selects a shipping carrier for an order.
/// </summary>
/// <remarks>
/// This executor uses <see cref="Executor{TInput}"/> (void return) combined with
/// <see cref="IWorkflowContext.SendMessageAsync"/> to forward the order to the next
/// connected executor (CreateShipment). This demonstrates explicit typed message passing
/// as an alternative to returning a value from the handler.
/// </remarks>
internal sealed class SelectCarrier() : Executor<OrderInfo>("SelectCarrier")
{
public override async ValueTask HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Blue;
Console.WriteLine($" [Shipping/SelectCarrier] Selecting carrier for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
message.Carrier = message.Amount > 50 ? "Express" : "Standard";
Console.ForegroundColor = ConsoleColor.Blue;
Console.WriteLine($" [Shipping/SelectCarrier] ✓ Selected carrier: {message.Carrier}");
Console.ResetColor();
// Use SendMessageAsync to forward the updated order to connected executors.
// With a void-return executor, this is the mechanism for passing data downstream.
await context.SendMessageAsync(message, cancellationToken: cancellationToken);
}
}
/// <summary>
/// Creates shipment and generates tracking number.
/// </summary>
internal sealed class CreateShipment() : Executor<OrderInfo, OrderInfo>("CreateShipment")
{
public override async ValueTask<OrderInfo> HandleAsync(OrderInfo message, IWorkflowContext context, CancellationToken cancellationToken = default)
{
Console.ForegroundColor = ConsoleColor.Blue;
Console.WriteLine($" [Shipping/CreateShipment] Creating shipment for order '{message.OrderId}'...");
Console.ResetColor();
await Task.Delay(TimeSpan.FromMilliseconds(100), cancellationToken);
message.TrackingNumber = $"TRACK-{Guid.NewGuid().ToString("N")[..10].ToUpperInvariant()}";
Console.ForegroundColor = ConsoleColor.Blue;
Console.WriteLine($" [Shipping/CreateShipment] ✓ Shipment created: {message.TrackingNumber}");
Console.ResetColor();
return message;
}
}
@@ -0,0 +1,146 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates nested sub-workflows. A sub-workflow can act as an executor
// within another workflow, including multi-level nesting (sub-workflow within sub-workflow).
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using SubWorkflows;
// Get DTS connection string from environment variable
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Build the FraudCheck sub-workflow (this will be nested inside the Payment sub-workflow)
AnalyzePatterns analyzePatterns = new();
CalculateRiskScore calculateRiskScore = new();
Workflow fraudCheckWorkflow = new WorkflowBuilder(analyzePatterns)
.WithName("SubFraudCheck")
.WithDescription("Analyzes transaction patterns and calculates risk score")
.AddEdge(analyzePatterns, calculateRiskScore)
.Build();
// Build the Payment sub-workflow: ValidatePayment -> FraudCheck (sub-workflow) -> ChargePayment
ValidatePayment validatePayment = new();
ExecutorBinding fraudCheckExecutor = fraudCheckWorkflow.BindAsExecutor("FraudCheck");
ChargePayment chargePayment = new();
Workflow paymentWorkflow = new WorkflowBuilder(validatePayment)
.WithName("SubPaymentProcessing")
.WithDescription("Validates and processes payment for an order")
.AddEdge(validatePayment, fraudCheckExecutor)
.AddEdge(fraudCheckExecutor, chargePayment)
.Build();
// Build the Shipping sub-workflow: SelectCarrier -> CreateShipment
SelectCarrier selectCarrier = new();
CreateShipment createShipment = new();
Workflow shippingWorkflow = new WorkflowBuilder(selectCarrier)
.WithName("SubShippingArrangement")
.WithDescription("Selects carrier and creates shipment")
.AddEdge(selectCarrier, createShipment)
.Build();
// Build the main workflow using sub-workflows as executors
// OrderReceived -> Payment (sub-workflow) -> Shipping (sub-workflow) -> OrderCompleted
OrderReceived orderReceived = new();
OrderCompleted orderCompleted = new();
ExecutorBinding paymentExecutor = paymentWorkflow.BindAsExecutor("Payment");
ExecutorBinding shippingExecutor = shippingWorkflow.BindAsExecutor("Shipping");
Workflow orderProcessingWorkflow = new WorkflowBuilder(orderReceived)
.WithName("OrderProcessing")
.WithDescription("Processes an order through payment and shipping")
.AddEdge(orderReceived, paymentExecutor)
.AddEdge(paymentExecutor, shippingExecutor)
.AddEdge(shippingExecutor, orderCompleted)
.Build();
// Configure and start the host
// Register only the main workflow - sub-workflows are discovered automatically!
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
workflowOptions => workflowOptions.AddWorkflow(orderProcessingWorkflow),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
Console.WriteLine("Durable Sub-Workflows Sample");
Console.WriteLine("Workflow: OrderReceived -> Payment(sub) -> Shipping(sub) -> OrderCompleted");
Console.WriteLine(" Payment contains nested FraudCheck sub-workflow (Level 2 nesting)");
Console.WriteLine();
Console.WriteLine("Enter an order ID (or 'exit'):");
while (true)
{
Console.Write("> ");
string? input = Console.ReadLine();
if (string.IsNullOrWhiteSpace(input) || input.Equals("exit", StringComparison.OrdinalIgnoreCase))
{
break;
}
try
{
await StartNewWorkflowAsync(input, orderProcessingWorkflow, workflowClient);
}
catch (Exception ex)
{
Console.WriteLine($"Error: {ex.Message}");
}
Console.WriteLine();
}
await host.StopAsync();
// Start a new workflow using streaming to observe events (including from sub-workflows)
static async Task StartNewWorkflowAsync(string orderId, Workflow workflow, IWorkflowClient client)
{
Console.WriteLine($"\nStarting order processing for '{orderId}'...");
IStreamingWorkflowRun run = await client.StreamAsync(workflow, orderId);
Console.WriteLine($"Run ID: {run.RunId}");
Console.WriteLine();
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
switch (evt)
{
// Custom event emitted from the FraudCheck sub-sub-workflow
case FraudRiskAssessedEvent e:
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine($" [Event from sub-workflow] {e.GetType().Name}: Risk score {e.RiskScore}/100");
Console.ResetColor();
break;
case DurableWorkflowCompletedEvent e:
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine($"✓ Order completed: {e.Result}");
Console.ResetColor();
break;
case DurableWorkflowFailedEvent e:
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine($"✗ Failed: {e.ErrorMessage}");
Console.ResetColor();
break;
}
}
}
@@ -0,0 +1,105 @@
# Sub-Workflows Sample (Nested Workflows)
This sample demonstrates how to compose complex workflows from simpler, reusable sub-workflows. Sub-workflows are built using `WorkflowBuilder` and embedded as executors via `BindAsExecutor()`. Unlike the in-process workflow runner, the durable workflow backend persists execution state across process restarts — each sub-workflow runs as a separate orchestration instance on the Durable Task Scheduler, providing independent checkpointing, fault tolerance, and hierarchical visualization in the DTS dashboard.
## Key Concepts Demonstrated
- **Sub-workflows**: Using `Workflow.BindAsExecutor()` to embed a workflow as an executor in another workflow
- **Multi-level nesting**: Sub-workflows within sub-workflows (Level 2 nesting)
- **Automatic discovery**: Registering only the main workflow; sub-workflows are discovered automatically
- **Failure isolation**: Each sub-workflow runs as a separate orchestration instance on the DTS backend
- **Hierarchical visualization**: Parent-child orchestration hierarchy visible in the DTS dashboard
- **Event propagation**: Custom workflow events (`FraudRiskAssessedEvent`) bubble up from nested sub-workflows to the streaming client
- **Message passing**: Using `Executor<TInput>` (void return) with `SendMessageAsync` to forward typed messages to connected executors (`SelectCarrier`)
- **Shared state within sub-workflows**: Using `QueueStateUpdateAsync`/`ReadStateAsync` to share data between executors within a sub-workflow (`AnalyzePatterns``CalculateRiskScore`)
## Overview
The sample implements an order processing workflow composed of two sub-workflows, one of which contains its own nested sub-workflow:
```
OrderProcessing (main workflow)
├── OrderReceived
├── Payment (sub-workflow)
│ ├── ValidatePayment
│ ├── FraudCheck (sub-sub-workflow) ← Level 2 nesting!
│ │ ├── AnalyzePatterns
│ │ └── CalculateRiskScore
│ └── ChargePayment
├── Shipping (sub-workflow)
│ ├── SelectCarrier ← Uses SendMessageAsync (void-return executor)
│ └── CreateShipment
└── OrderCompleted
```
| Executor | Sub-Workflow | Description |
|----------|-------------|-------------|
| OrderReceived | Main | Receives order ID and creates order info |
| ValidatePayment | Payment | Validates payment information |
| AnalyzePatterns | FraudCheck (nested in Payment) | Analyzes transaction patterns, stores results in shared state |
| CalculateRiskScore | FraudCheck (nested in Payment) | Reads shared state, calculates risk score, emits `FraudRiskAssessedEvent` |
| ChargePayment | Payment | Charges payment amount |
| SelectCarrier | Shipping | Selects carrier using `SendMessageAsync` (void-return executor) |
| CreateShipment | Shipping | Creates shipment with tracking |
| OrderCompleted | Main | Outputs completed order summary |
## How Sub-Workflows Work
For an introduction to sub-workflows and the `BindAsExecutor()` API, see the [Sub-Workflows foundational sample](../../../../03-workflows/_StartHere/05_SubWorkflows).
This durable sample extends the same pattern — the key difference is that each sub-workflow runs as a **separate orchestration instance** on the Durable Task Scheduler, providing independent checkpointing, fault tolerance, and hierarchical visualization in the DTS dashboard.
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on configuring the environment, including how to install and run the Durable Task Scheduler.
## Running the Sample
```bash
cd dotnet/samples/04-hosting/DurableWorkflows/ConsoleApps/07_SubWorkflows
dotnet run --framework net10.0
```
### Sample Output
```text
Durable Sub-Workflows Sample
Workflow: OrderReceived -> Payment(sub) -> Shipping(sub) -> OrderCompleted
Payment contains nested FraudCheck sub-workflow (Level 2 nesting)
Enter an order ID (or 'exit'):
> ORD-001
Starting order processing for 'ORD-001'...
Run ID: abc123...
[OrderReceived] Processing order 'ORD-001'
[Payment/ValidatePayment] Validating payment for order 'ORD-001'...
[Payment/ValidatePayment] Payment validated for $99.99
[Payment/FraudCheck/AnalyzePatterns] Analyzing patterns for order 'ORD-001'...
[Payment/FraudCheck/AnalyzePatterns] ✓ Pattern analysis complete (2 suspicious patterns)
[Payment/FraudCheck/CalculateRiskScore] Calculating risk score for order 'ORD-001'...
[Payment/FraudCheck/CalculateRiskScore] ✓ Risk score: 53/100 (based on 2 patterns)
[Event from sub-workflow] FraudRiskAssessedEvent: Risk score 53/100
[Payment/ChargePayment] Charging $99.99 for order 'ORD-001'...
[Payment/ChargePayment] ✓ Payment processed: TXN-A1B2C3D4
[Shipping/SelectCarrier] Selecting carrier for order 'ORD-001'...
[Shipping/SelectCarrier] ✓ Selected carrier: Express
[Shipping/CreateShipment] Creating shipment for order 'ORD-001'...
[Shipping/CreateShipment] ✓ Shipment created: TRACK-I9J0K1L2M3
┌─────────────────────────────────────────────────────────────────┐
│ [OrderCompleted] Order 'ORD-001' successfully processed!
│ Payment: TXN-A1B2C3D4
│ Shipping: Express - TRACK-I9J0K1L2M3
└─────────────────────────────────────────────────────────────────┘
✓ Order completed: Order ORD-001 completed. Tracking: TRACK-I9J0K1L2M3
> exit
```
### Viewing Workflows in the DTS Dashboard
After running the workflow, you can navigate to the Durable Task Scheduler (DTS) dashboard to inspect the orchestration hierarchy, including sub-orchestrations.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
Because each sub-workflow runs as a separate orchestration instance, the dashboard shows a parent-child hierarchy: the top-level `OrderProcessing` orchestration with `Payment` and `Shipping` as child orchestrations, and `FraudCheck` nested under `Payment`. You can click into each orchestration to inspect its executor inputs/outputs, events, and execution timeline independently.
@@ -0,0 +1,28 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFrameworks>net10.0</TargetFrameworks>
<OutputType>Exe</OutputType>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
<AssemblyName>WorkflowHITL</AssemblyName>
<RootNamespace>WorkflowHITL</RootNamespace>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="Azure.Identity" />
<PackageReference Include="Microsoft.DurableTask.Client.AzureManaged" />
<PackageReference Include="Microsoft.DurableTask.Worker.AzureManaged" />
<PackageReference Include="Microsoft.Extensions.Hosting" />
</ItemGroup>
<!-- Local projects that should be switched to package references when using the sample outside of this MAF repo -->
<!--
<ItemGroup>
<PackageReference Include="Microsoft.Agents.AI.DurableTask" />
<PackageReference Include="Microsoft.Agents.AI.Workflows" />
</ItemGroup>
-->
<ItemGroup>
<ProjectReference Include="..\..\..\..\..\src\Microsoft.Agents.AI.DurableTask\Microsoft.Agents.AI.DurableTask.csproj" />
</ItemGroup>
</Project>
@@ -0,0 +1,81 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace WorkflowHITL;
/// <summary>
/// Represents an expense approval request.
/// </summary>
/// <param name="ExpenseId">The unique identifier of the expense.</param>
/// <param name="Amount">The amount of the expense.</param>
/// <param name="EmployeeName">The name of the employee submitting the expense.</param>
public record ApprovalRequest(string ExpenseId, decimal Amount, string EmployeeName);
/// <summary>
/// Represents the response to an approval request.
/// </summary>
/// <param name="Approved">Whether the expense was approved.</param>
/// <param name="Comments">Optional comments from the approver.</param>
public record ApprovalResponse(bool Approved, string? Comments);
/// <summary>
/// Retrieves expense details and creates an approval request.
/// </summary>
internal sealed class CreateApprovalRequest() : Executor<string, ApprovalRequest>("RetrieveRequest")
{
/// <inheritdoc/>
public override ValueTask<ApprovalRequest> HandleAsync(
string message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// In a real scenario, this would look up expense details from a database
return new ValueTask<ApprovalRequest>(new ApprovalRequest(message, 1500.00m, "Jerry"));
}
}
/// <summary>
/// Prepares the approval request for finance review after manager approval.
/// </summary>
internal sealed class PrepareFinanceReview() : Executor<ApprovalResponse, ApprovalRequest>("PrepareFinanceReview")
{
/// <inheritdoc/>
public override ValueTask<ApprovalRequest> HandleAsync(
ApprovalResponse message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
if (!message.Approved)
{
throw new InvalidOperationException("Cannot proceed to finance review — manager denied the expense.");
}
// In a real scenario, this would retrieve the original expense details
return new ValueTask<ApprovalRequest>(new ApprovalRequest("EXP-2025-001", 1500.00m, "Jerry"));
}
}
/// <summary>
/// Processes the expense reimbursement based on the parallel approval responses from budget and compliance.
/// </summary>
internal sealed class ExpenseReimburse() : Executor<ApprovalResponse[], string>("Reimburse")
{
/// <inheritdoc/>
public override async ValueTask<string> HandleAsync(
ApprovalResponse[] message,
IWorkflowContext context,
CancellationToken cancellationToken = default)
{
// Check that all parallel approvals passed
ApprovalResponse? denied = Array.Find(message, r => !r.Approved);
if (denied is not null)
{
return $"Expense reimbursement denied. Comments: {denied.Comments}";
}
// Simulate payment processing
await Task.Delay(1000, cancellationToken);
return $"Expense reimbursed at {DateTime.UtcNow:O}";
}
}
@@ -0,0 +1,98 @@
// Copyright (c) Microsoft. All rights reserved.
// This sample demonstrates a Human-in-the-Loop (HITL) workflow using Durable Tasks.
//
// ┌──────────────────────┐ ┌────────────────┐ ┌─────────────────────┐ ┌────────────────────┐
// │ CreateApprovalRequest│──►│ManagerApproval │──►│PrepareFinanceReview │──┬►│ BudgetApproval │──┐
// └──────────────────────┘ │ (RequestPort) │ └─────────────────────┘ │ │ (RequestPort) │ │
// └────────────────┘ │ └────────────────────┘ │ ┌─────────────────┐
// │ ├─►│ExpenseReimburse │
// │ ┌────────────────────┐ │ └─────────────────┘
// └►│ComplianceApproval │──┘
// │ (RequestPort) │
// └────────────────────┘
//
// The workflow pauses at three RequestPorts — one for the manager, then two in parallel for finance.
// After manager approval, BudgetApproval and ComplianceApproval run concurrently via fan-out/fan-in.
using Microsoft.Agents.AI.DurableTask;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask.Client.AzureManaged;
using Microsoft.DurableTask.Worker.AzureManaged;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using WorkflowHITL;
string dtsConnectionString = Environment.GetEnvironmentVariable("DURABLE_TASK_SCHEDULER_CONNECTION_STRING")
?? "Endpoint=http://localhost:8080;TaskHub=default;Authentication=None";
// Define executors and RequestPorts for the three HITL pause points
CreateApprovalRequest createRequest = new();
RequestPort<ApprovalRequest, ApprovalResponse> managerApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("ManagerApproval");
PrepareFinanceReview prepareFinanceReview = new();
RequestPort<ApprovalRequest, ApprovalResponse> budgetApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("BudgetApproval");
RequestPort<ApprovalRequest, ApprovalResponse> complianceApproval = RequestPort.Create<ApprovalRequest, ApprovalResponse>("ComplianceApproval");
ExpenseReimburse reimburse = new();
// Build the workflow: CreateApprovalRequest -> ManagerApproval -> PrepareFinanceReview -> [BudgetApproval AND ComplianceApproval] -> ExpenseReimburse
Workflow expenseApproval = new WorkflowBuilder(createRequest)
.WithName("ExpenseReimbursement")
.WithDescription("Expense reimbursement with manager and parallel finance approvals")
.AddEdge(createRequest, managerApproval)
.AddEdge(managerApproval, prepareFinanceReview)
.AddFanOutEdge(prepareFinanceReview, [budgetApproval, complianceApproval])
.AddFanInBarrierEdge([budgetApproval, complianceApproval], reimburse)
.Build();
IHost host = Host.CreateDefaultBuilder(args)
.ConfigureLogging(logging => logging.SetMinimumLevel(LogLevel.Warning))
.ConfigureServices(services =>
{
services.ConfigureDurableWorkflows(
options => options.AddWorkflow(expenseApproval),
workerBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString),
clientBuilder: builder => builder.UseDurableTaskScheduler(dtsConnectionString));
})
.Build();
await host.StartAsync();
IWorkflowClient workflowClient = host.Services.GetRequiredService<IWorkflowClient>();
// Start the workflow with streaming to observe events including HITL pauses
string expenseId = "EXP-2025-001";
Console.WriteLine($"Starting expense reimbursement workflow for expense: {expenseId}");
IStreamingWorkflowRun run = await workflowClient.StreamAsync(expenseApproval, expenseId);
Console.WriteLine($"Workflow started with instance ID: {run.RunId}\n");
// Watch for workflow events — handle HITL requests as they arrive
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
switch (evt)
{
case DurableWorkflowWaitingForInputEvent requestEvent:
Console.WriteLine($"Workflow paused at RequestPort: {requestEvent.RequestPort.Id}");
Console.WriteLine($" Input: {requestEvent.Input}");
// In a real scenario, this would involve human interaction (UI, email, Teams, etc.)
ApprovalRequest? request = requestEvent.GetInputAs<ApprovalRequest>();
Console.WriteLine($" Approval for: {request?.EmployeeName}, Amount: {request?.Amount:C}");
ApprovalResponse approvalResponse = new(Approved: true, Comments: "Approved by manager.");
await run.SendResponseAsync(requestEvent, approvalResponse);
Console.WriteLine($" Response sent: Approved={approvalResponse.Approved}\n");
break;
case DurableWorkflowCompletedEvent completedEvent:
Console.WriteLine($"Workflow completed: {completedEvent.Result}");
break;
case DurableWorkflowFailedEvent failedEvent:
Console.WriteLine($"Workflow failed: {failedEvent.ErrorMessage}");
break;
}
}
await host.StopAsync();
@@ -0,0 +1,106 @@
# Workflow Human-in-the-Loop (HITL) Sample
This sample demonstrates a **Human-in-the-Loop** pattern in durable workflows using `RequestPort`. The workflow pauses execution at a manager approval point, then fans out to two parallel finance approval points — budget and compliance — before resuming.
## Key Concepts Demonstrated
- Using `RequestPort` to define external input points in a workflow
- Sequential and parallel HITL pause points in a single workflow using fan-out/fan-in
- Streaming workflow events with `IStreamingWorkflowRun`
- Handling `DurableWorkflowWaitingForInputEvent` to detect HITL pauses
- Using `SendResponseAsync` to provide responses and resume the workflow
- **Durability**: The workflow survives process restarts while waiting for human input
## Workflow
This sample implements the following workflow:
```
┌──────────────────────┐ ┌────────────────┐ ┌─────────────────────┐ ┌────────────────────┐
│ CreateApprovalRequest│──►│ManagerApproval │──►│PrepareFinanceReview │──┬►│ BudgetApproval │──┐
└──────────────────────┘ │ (RequestPort) │ └─────────────────────┘ │ │ (RequestPort) │ │
└────────────────┘ │ └────────────────────┘ │ ┌─────────────────┐
│ ├─►│ExpenseReimburse │
│ ┌────────────────────┐ │ └─────────────────┘
└►│ComplianceApproval │──┘
│ (RequestPort) │
└────────────────────┘
```
| Step | Description |
|------|-------------|
| CreateApprovalRequest | Retrieves expense details and creates an approval request |
| ManagerApproval (RequestPort) | **PAUSES** the workflow and waits for manager approval |
| PrepareFinanceReview | Prepares the request for finance review after manager approval |
| BudgetApproval (RequestPort) | **PAUSES** the workflow and waits for budget approval (parallel) |
| ComplianceApproval (RequestPort) | **PAUSES** the workflow and waits for compliance approval (parallel) |
| ExpenseReimburse | Processes the reimbursement after all approvals pass |
## How It Works
A `RequestPort` defines a typed external input point in the workflow:
```csharp
RequestPort<ApprovalRequest, ApprovalResponse> managerApproval =
RequestPort.Create<ApprovalRequest, ApprovalResponse>("ManagerApproval");
```
Use `WatchStreamAsync` to observe events. When the workflow reaches a `RequestPort`, a `DurableWorkflowWaitingForInputEvent` is emitted. Call `SendResponseAsync` to provide the response and resume the workflow:
```csharp
await foreach (WorkflowEvent evt in run.WatchStreamAsync())
{
switch (evt)
{
case DurableWorkflowWaitingForInputEvent requestEvent:
ApprovalRequest? request = requestEvent.GetInputAs<ApprovalRequest>();
await run.SendResponseAsync(requestEvent, new ApprovalResponse(Approved: true, Comments: "Approved."));
break;
}
}
```
## Environment Setup
See the [README.md](../../README.md) file in the parent directory for information on configuring the environment, including how to install and run the Durable Task Scheduler.
## Running the Sample
```bash
cd dotnet/samples/04-hosting/DurableWorkflows/ConsoleApps/08_WorkflowHITL
dotnet run --framework net10.0
```
### Sample Output
```text
Starting expense reimbursement workflow for expense: EXP-2025-001
Workflow started with instance ID: abc123...
Workflow paused at RequestPort: ManagerApproval
Input: {"expenseId":"EXP-2025-001","amount":1500.00,"employeeName":"Jerry"}
Approval for: Jerry, Amount: $1,500.00
Response sent: Approved=True
Workflow paused at RequestPort: BudgetApproval
Input: {"expenseId":"EXP-2025-001","amount":1500.00,"employeeName":"Jerry"}
Approval for: Jerry, Amount: $1,500.00
Response sent: Approved=True
Workflow paused at RequestPort: ComplianceApproval
Input: {"expenseId":"EXP-2025-001","amount":1500.00,"employeeName":"Jerry"}
Approval for: Jerry, Amount: $1,500.00
Response sent: Approved=True
Workflow completed: Expense reimbursed at 2025-01-23T17:30:00.0000000Z
```
### Viewing Workflows in the DTS Dashboard
After running the sample, you can navigate to the Durable Task Scheduler (DTS) dashboard to visualize the completed orchestration and inspect its execution history.
If you are using the DTS emulator, the dashboard is available at `http://localhost:8082`.
1. Open the dashboard and look for the orchestration instance matching the instance ID logged in the console output (e.g., `abc123...`).
2. Click into the instance to see the execution timeline, which shows each executor activity and the `WaitForExternalEvent` pauses where the workflow waited for human input — including the two parallel finance approvals.
3. Expand individual activity steps to inspect inputs and outputs — for example, the `ManagerApproval`, `BudgetApproval`, and `ComplianceApproval` external events will show the approval request sent and the response received.
@@ -0,0 +1,5 @@
<Project> <Import Project="../../Directory.Build.props" /> <!-- Remove the Environment alias from parent Directory.Build.props to allow System.Environment usage -->
<ItemGroup>
<Using Remove="SampleHelpers.SampleEnvironment" />
</ItemGroup>
</Project>
@@ -0,0 +1,50 @@
# Durable Workflow Samples
This directory contains samples demonstrating how to build durable workflows using the Microsoft Agent Framework.
## Environment Setup
### Prerequisites
- [.NET 10 SDK](https://dotnet.microsoft.com/download/dotnet/10.0) or later
- [Durable Task Scheduler](https://learn.microsoft.com/en-us/azure/azure-functions/durable/durable-task-scheduler/durable-task-scheduler) running locally or in Azure
### Running the Durable Task Scheduler Emulator
To run the emulator locally using Docker:
```bash
docker run -d -p 8080:8080 --name durabletask-emulator mcr.microsoft.com/durabletask/emulator:latest
```
Set the connection string environment variable to point to the local emulator:
```bash
# Linux/macOS
export DURABLE_TASK_SCHEDULER_CONNECTION_STRING="AccountEndpoint=http://localhost:8080"
# Windows (PowerShell)
$env:DURABLE_TASK_SCHEDULER_CONNECTION_STRING = "AccountEndpoint=http://localhost:8080"
```
## Samples
### Console Apps
| Sample | Description |
|--------|-------------|
| [01_SequentialWorkflow](ConsoleApps/01_SequentialWorkflow/) | Basic sequential workflow with ordered executor steps |
| [02_ConcurrentWorkflow](ConsoleApps/02_ConcurrentWorkflow/) | Fan-out/fan-in concurrent workflow execution |
| [03_ConditionalEdges](ConsoleApps/03_ConditionalEdges/) | Workflows with conditional routing between executors |
| [05_WorkflowEvents](ConsoleApps/05_WorkflowEvents/) | Publishing and subscribing to workflow events |
| [06_WorkflowSharedState](ConsoleApps/06_WorkflowSharedState/) | Sharing state across workflow executors |
| [07_SubWorkflows](ConsoleApps/07_SubWorkflows/) | Nested sub-workflow composition |
| [08_WorkflowHITL](ConsoleApps/08_WorkflowHITL/) | Human-in-the-loop workflow with approval gates |
### Azure Functions
| Sample | Description |
|--------|-------------|
| [01_SequentialWorkflow](AzureFunctions/01_SequentialWorkflow/) | Sequential workflow hosted in Azure Functions |
| [02_ConcurrentWorkflow](AzureFunctions/02_ConcurrentWorkflow/) | Concurrent workflow hosted in Azure Functions |
| [03_WorkflowHITL](AzureFunctions/03_WorkflowHITL/) | Human-in-the-loop workflow hosted in Azure Functions |
@@ -15,7 +15,6 @@
<ItemGroup>
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Hosting.AGUI.AspNetCore\Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.AGUI\Microsoft.Agents.AI.AGUI.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.OpenAI\Microsoft.Agents.AI.OpenAI.csproj" />
</ItemGroup>
@@ -9,10 +9,12 @@
<ItemGroup>
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Hosting.A2A\Microsoft.Agents.AI.Hosting.A2A.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Hosting.OpenAI\Microsoft.Agents.AI.Hosting.OpenAI.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Hosting\Microsoft.Agents.AI.Hosting.csproj" />
<ProjectReference Include="..\AgentWebChat.ServiceDefaults\AgentWebChat.ServiceDefaults.csproj" />
<ProjectReference Include="..\..\..\..\src\Microsoft.Agents.AI.Abstractions\Microsoft.Agents.AI.Abstractions.csproj" />
</ItemGroup>
<ItemGroup>
<PackageReference Include="OpenAI" />
<PackageReference Include="Microsoft.Extensions.AI.OpenAI" />
</ItemGroup>
</Project>
+9
View File
@@ -0,0 +1,9 @@
<Project>
<Import Project="../Directory.Build.props" />
<ItemGroup>
<PackageReference Include="ReferenceTrimmer" PrivateAssets="all" IncludeAssets="build;analyzers;buildTransitive" />
</ItemGroup>
</Project>
@@ -16,12 +16,9 @@
<Description>Provides Microsoft Agent Framework support for Agent-User Interaction (AG-UI) protocol client functionality.</Description>
</PropertyGroup>
<ItemGroup>
<ProjectReference Include="..\Microsoft.Agents.AI.Abstractions\Microsoft.Agents.AI.Abstractions.csproj" />
</ItemGroup>
<ItemGroup>
<PackageReference Include="Microsoft.Extensions.AI" />
<PackageReference Include="Microsoft.Extensions.Logging.Abstractions" />
<PackageReference Include="System.Net.ServerSentEvents" />
<PackageReference Include="System.Net.Http.Json" />
<PackageReference Include="System.Threading.Channels" />
@@ -28,7 +28,6 @@
<ItemGroup>
<PackageReference Include="Microsoft.Extensions.AI.Abstractions" />
<PackageReference Include="Microsoft.Extensions.Logging.Abstractions" />
</ItemGroup>
<ItemGroup>
@@ -2,7 +2,36 @@
## [Unreleased]
### Changed
- Added support for durable workflows ([#4436](https://github.com/microsoft/agent-framework/pull/4436))
## v1.0.0-preview.260219.1
- [BREAKING] Changed ChatHistory and AIContext Providers to have pipeline semantics ([#3806](https://github.com/microsoft/agent-framework/pull/3806))
- Marked all `RunAsync<T>` overloads as `new`, added missing ones, and added support for primitives and arrays ([#3803](https://github.com/microsoft/agent-framework/pull/3803))
- Improve session cast error message quality and consistency ([#3973](https://github.com/microsoft/agent-framework/pull/3973))
## v1.0.0-preview.260212.1
- [BREAKING] Changed AIAgent.SerializeSession to AIAgent.SerializeSessionAsync ([#3879](https://github.com/microsoft/agent-framework/pull/3879))
## v1.0.0-preview.260209.1
- [BREAKING] Introduce Core method pattern for Session management methods on AIAgent ([#3699](https://github.com/microsoft/agent-framework/pull/3699))
## v1.0.0-preview.260205.1
- [BREAKING] Moved AgentSession.Serialize to AIAgent.SerializeSession ([#3650](https://github.com/microsoft/agent-framework/pull/3650))
- [BREAKING] Renamed serializedSession parameter to serializedState on DeserializeSessionAsync for consistency ([#3681](https://github.com/microsoft/agent-framework/pull/3681))
## v1.0.0-preview.260127.1
- [BREAKING] Renamed AgentThread to AgentSession ([#3430](https://github.com/microsoft/agent-framework/pull/3430))
## v1.0.0-preview.260108.1
- [BREAKING] Removed AgentThreadMetadata and used AgentSessionId directly instead ([#3067](https://github.com/microsoft/agent-framework/pull/3067))
## v1.0.0-preview.251219.1
- Filter empty `AIContent` from durable agent state responses ([#4670](https://github.com/microsoft/agent-framework/pull/4670))
@@ -12,15 +41,6 @@
- Added TTL configuration for durable agent entities ([#2679](https://github.com/microsoft/agent-framework/pull/2679))
- Switch to new "Run" method name ([#2843](https://github.com/microsoft/agent-framework/pull/2843))
- Removed AgentThreadMetadata and used AgentSessionId directly instead ([#3067](https://github.com/microsoft/agent-framework/pull/3067));
- Renamed AgentThread to AgentSession ([#3430](https://github.com/microsoft/agent-framework/pull/3430))
- Moved AgentSession.Serialize to AIAgent.SerializeSession ([#3650](https://github.com/microsoft/agent-framework/pull/3650))
- Renamed serializedSession parameter to serializedState on DeserializeSessionAsync for consistency ([#3681](https://github.com/microsoft/agent-framework/pull/3681))
- Introduce Core method pattern for Session management methods on AIAgent ([#3699](https://github.com/microsoft/agent-framework/pull/3699))
- Changed AIAgent.SerializeSession to AIAgent.SerializeSessionAsync ([#3879](https://github.com/microsoft/agent-framework/pull/3879))
- Changed ChatHistory and AIContext Providers to have pipeline semantics ([#3806](https://github.com/microsoft/agent-framework/pull/3806))
- Marked all `RunAsync<T>` overloads as `new`, added missing ones, and added support for primitives and arrays ([#3803](https://github.com/microsoft/agent-framework/pull/3803))
- Improve session cast error message quality and consistency ([#3973](https://github.com/microsoft/agent-framework/pull/3973))
NOTE: Some of the above changes may have been part of earlier releases not mentioned in this file.
@@ -141,4 +141,15 @@ public sealed class DurableAgentsOptions
{
return this._agentTimeToLive.TryGetValue(agentName, out TimeSpan? ttl) ? ttl : this.DefaultTimeToLive;
}
/// <summary>
/// Determines whether an agent with the specified name is registered.
/// </summary>
/// <param name="agentName">The name of the agent to locate. Cannot be null.</param>
/// <returns>true if an agent with the specified name is registered; otherwise, false.</returns>
internal bool ContainsAgent(string agentName)
{
ArgumentNullException.ThrowIfNull(agentName);
return this._agentFactories.ContainsKey(agentName);
}
}
@@ -0,0 +1,66 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics.CodeAnalysis;
using System.Text.Json;
using System.Text.Json.Serialization.Metadata;
using Microsoft.Agents.AI.DurableTask.State;
using Microsoft.DurableTask;
namespace Microsoft.Agents.AI.DurableTask;
/// <summary>
/// Custom data converter for durable agents and workflows that ensures proper JSON serialization.
/// </summary>
/// <remarks>
/// This converter handles special cases like <see cref="DurableAgentState"/> using source-generated
/// JSON contexts for AOT compatibility, and falls back to reflection-based serialization for other types.
/// </remarks>
internal sealed class DurableDataConverter : DataConverter
{
private static readonly JsonSerializerOptions s_options = new(DurableAgentJsonUtilities.DefaultOptions)
{
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
PropertyNameCaseInsensitive = true,
};
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Fallback uses reflection when metadata unavailable.")]
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Fallback uses reflection when metadata unavailable.")]
public override object? Deserialize(string? data, Type targetType)
{
if (data is null)
{
return null;
}
if (targetType == typeof(DurableAgentState))
{
return JsonSerializer.Deserialize(data, DurableAgentStateJsonContext.Default.DurableAgentState);
}
JsonTypeInfo? typeInfo = s_options.GetTypeInfo(targetType);
return typeInfo is not null
? JsonSerializer.Deserialize(data, typeInfo)
: JsonSerializer.Deserialize(data, targetType, s_options);
}
[return: NotNullIfNotNull(nameof(value))]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Fallback uses reflection when metadata unavailable.")]
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Fallback uses reflection when metadata unavailable.")]
public override string? Serialize(object? value)
{
if (value is null)
{
return null;
}
if (value is DurableAgentState durableAgentState)
{
return JsonSerializer.Serialize(durableAgentState, DurableAgentStateJsonContext.Default.DurableAgentState);
}
JsonTypeInfo? typeInfo = s_options.GetTypeInfo(value.GetType());
return typeInfo is not null
? JsonSerializer.Serialize(value, typeInfo)
: JsonSerializer.Serialize(value, s_options);
}
}
@@ -0,0 +1,31 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics;
using Microsoft.Agents.AI.DurableTask.Workflows;
namespace Microsoft.Agents.AI.DurableTask;
/// <summary>
/// Provides configuration options for durable agents and workflows.
/// </summary>
[DebuggerDisplay("Workflows = {Workflows.Workflows.Count}, Agents = {Agents.AgentCount}")]
public class DurableOptions
{
/// <summary>
/// Initializes a new instance of the <see cref="DurableOptions"/> class.
/// </summary>
internal DurableOptions()
{
this.Workflows = new DurableWorkflowOptions(this);
}
/// <summary>
/// Gets the configuration options for durable agents.
/// </summary>
public DurableAgentsOptions Agents { get; } = new();
/// <summary>
/// Gets the configuration options for durable workflows.
/// </summary>
public DurableWorkflowOptions Workflows { get; }
}
@@ -0,0 +1,34 @@
// Copyright (c) Microsoft. All rights reserved.
namespace Microsoft.Agents.AI.DurableTask;
/// <summary>
/// Marker class used to track whether core durable task services have been registered.
/// </summary>
/// <remarks>
/// <para>
/// <b>Problem it solves:</b> Users may call configuration methods multiple times:
/// <code>
/// services.ConfigureDurableOptions(...); // 1st call - registers agent A
/// services.ConfigureDurableOptions(...); // 2nd call - registers workflow X
/// services.ConfigureDurableOptions(...); // 3rd call - registers agent B and workflow Y
/// </code>
/// Each call invokes <c>EnsureDurableServicesRegistered</c>. Without this marker, core services like
/// <c>AddDurableTaskWorker</c> and <c>AddDurableTaskClient</c> would be registered multiple times,
/// causing runtime errors or unexpected behavior.
/// </para>
/// <para>
/// <b>How it works:</b>
/// <list type="number">
/// <item><description>First call: No marker in services → register marker + all core services</description></item>
/// <item><description>Subsequent calls: Marker exists → early return, skip core service registration</description></item>
/// </list>
/// </para>
/// <para>
/// <b>Why not use TryAddSingleton for everything?</b>
/// While <c>TryAddSingleton</c> prevents duplicate simple service registrations, it doesn't work for
/// complex registrations like <c>AddDurableTaskWorker</c> which have side effects and configure
/// internal builders. The marker pattern provides a clean, explicit guard for the entire registration block.
/// </para>
/// </remarks>
internal sealed class DurableServicesMarker;
@@ -100,4 +100,131 @@ internal static partial class Logs
public static partial void LogTTLExpirationTimeCleared(
this ILogger logger,
AgentSessionId sessionId);
// Durable workflow logs (EventIds 100-199)
[LoggerMessage(
EventId = 100,
Level = LogLevel.Information,
Message = "Starting workflow '{WorkflowName}' with instance '{InstanceId}'")]
public static partial void LogWorkflowStarting(
this ILogger logger,
string workflowName,
string instanceId);
[LoggerMessage(
EventId = 101,
Level = LogLevel.Information,
Message = "Superstep {Step}: {Count} active executor(s)")]
public static partial void LogSuperstepStarting(
this ILogger logger,
int step,
int count);
[LoggerMessage(
EventId = 102,
Level = LogLevel.Debug,
Message = "Superstep {Step} executors: [{Executors}]")]
public static partial void LogSuperstepExecutors(
this ILogger logger,
int step,
string executors);
[LoggerMessage(
EventId = 103,
Level = LogLevel.Information,
Message = "Workflow completed")]
public static partial void LogWorkflowCompleted(
this ILogger logger);
[LoggerMessage(
EventId = 104,
Level = LogLevel.Warning,
Message = "Workflow '{InstanceId}' terminated early: reached maximum superstep limit ({MaxSupersteps}) with {RemainingExecutors} executor(s) still queued")]
public static partial void LogWorkflowMaxSuperstepsExceeded(
this ILogger logger,
string instanceId,
int maxSupersteps,
int remainingExecutors);
[LoggerMessage(
EventId = 105,
Level = LogLevel.Debug,
Message = "Fan-In executor {ExecutorId}: aggregated {Count} messages from [{Sources}]")]
public static partial void LogFanInAggregated(
this ILogger logger,
string executorId,
int count,
string sources);
[LoggerMessage(
EventId = 106,
Level = LogLevel.Debug,
Message = "Executor '{ExecutorId}' returned result (length: {Length}, messages: {MessageCount})")]
public static partial void LogExecutorResultReceived(
this ILogger logger,
string executorId,
int length,
int messageCount);
[LoggerMessage(
EventId = 107,
Level = LogLevel.Debug,
Message = "Dispatching executor '{ExecutorId}' (agentic: {IsAgentic})")]
public static partial void LogDispatchingExecutor(
this ILogger logger,
string executorId,
bool isAgentic);
[LoggerMessage(
EventId = 108,
Level = LogLevel.Warning,
Message = "Agent '{AgentName}' not found")]
public static partial void LogAgentNotFound(
this ILogger logger,
string agentName);
[LoggerMessage(
EventId = 109,
Level = LogLevel.Debug,
Message = "Edge {Source} -> {Sink}: condition returned false, skipping")]
public static partial void LogEdgeConditionFalse(
this ILogger logger,
string source,
string sink);
[LoggerMessage(
EventId = 110,
Level = LogLevel.Warning,
Message = "Failed to evaluate condition for edge {Source} -> {Sink}, skipping")]
public static partial void LogEdgeConditionEvaluationFailed(
this ILogger logger,
Exception ex,
string source,
string sink);
[LoggerMessage(
EventId = 111,
Level = LogLevel.Debug,
Message = "Edge {Source} -> {Sink}: routing message")]
public static partial void LogEdgeRoutingMessage(
this ILogger logger,
string source,
string sink);
[LoggerMessage(
EventId = 112,
Level = LogLevel.Information,
Message = "Workflow waiting for external input at RequestPort '{RequestPortId}'")]
public static partial void LogWaitingForExternalEvent(
this ILogger logger,
string requestPortId);
[LoggerMessage(
EventId = 113,
Level = LogLevel.Information,
Message = "Received external event for RequestPort '{RequestPortId}'")]
public static partial void LogReceivedExternalEvent(
this ILogger logger,
string requestPortId);
}
@@ -17,7 +17,6 @@
</PropertyGroup>
<PropertyGroup>
<InjectSharedThrow>true</InjectSharedThrow>
<InjectSharedStructuredOutput>true</InjectSharedStructuredOutput>
</PropertyGroup>
@@ -28,6 +27,7 @@
</ItemGroup>
<ItemGroup>
<ProjectReference Include="..\Microsoft.Agents.AI.Workflows\Microsoft.Agents.AI.Workflows.csproj" />
<ProjectReference Include="..\Microsoft.Agents.AI\Microsoft.Agents.AI.csproj" />
</ItemGroup>
@@ -1,18 +1,18 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics.CodeAnalysis;
using System.Text.Json;
using System.Text.Json.Serialization.Metadata;
using Microsoft.Agents.AI.DurableTask.State;
using Microsoft.Agents.AI.DurableTask.Workflows;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask;
using Microsoft.DurableTask.Client;
using Microsoft.DurableTask.Worker;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.DependencyInjection.Extensions;
using Microsoft.Extensions.Logging;
namespace Microsoft.Agents.AI.DurableTask;
/// <summary>
/// Agent-specific extension methods for the <see cref="IServiceCollection"/> class.
/// Extension methods for configuring durable agents and workflows with dependency injection.
/// </summary>
public static class ServiceCollectionExtensions
{
@@ -30,77 +30,331 @@ public static class ServiceCollectionExtensions
}
/// <summary>
/// Configures the Durable Agents services via the service collection.
/// Configures durable agents, automatically registering agent entities.
/// </summary>
/// <remarks>
/// <para>
/// This method provides an agent-focused configuration experience.
/// If you need to configure both agents and workflows, consider using
/// <see cref="ConfigureDurableOptions"/> instead.
/// </para>
/// <para>
/// Multiple calls to this method are supported and configurations are composed additively.
/// </para>
/// </remarks>
/// <param name="services">The service collection.</param>
/// <param name="configure">A delegate to configure the durable agents.</param>
/// <param name="workerBuilder">A delegate to configure the Durable Task worker.</param>
/// <param name="clientBuilder">A delegate to configure the Durable Task client.</param>
/// <returns>The service collection.</returns>
/// <param name="workerBuilder">Optional delegate to configure the Durable Task worker.</param>
/// <param name="clientBuilder">Optional delegate to configure the Durable Task client.</param>
/// <returns>The service collection for chaining.</returns>
public static IServiceCollection ConfigureDurableAgents(
this IServiceCollection services,
Action<DurableAgentsOptions> configure,
Action<IDurableTaskWorkerBuilder>? workerBuilder = null,
Action<IDurableTaskClientBuilder>? clientBuilder = null)
{
return services.ConfigureDurableOptions(
options => configure(options.Agents),
workerBuilder,
clientBuilder);
}
/// <summary>
/// Configures durable workflows, automatically registering orchestrations and activities.
/// </summary>
/// <remarks>
/// <para>
/// This method provides a workflow-focused configuration experience.
/// If you need to configure both agents and workflows, consider using
/// <see cref="ConfigureDurableOptions"/> instead.
/// </para>
/// <para>
/// Multiple calls to this method are supported and configurations are composed additively.
/// </para>
/// </remarks>
/// <param name="services">The service collection to configure.</param>
/// <param name="configure">A delegate to configure the workflow options.</param>
/// <param name="workerBuilder">Optional delegate to configure the durable task worker.</param>
/// <param name="clientBuilder">Optional delegate to configure the durable task client.</param>
/// <returns>The service collection for chaining.</returns>
public static IServiceCollection ConfigureDurableWorkflows(
this IServiceCollection services,
Action<DurableWorkflowOptions> configure,
Action<IDurableTaskWorkerBuilder>? workerBuilder = null,
Action<IDurableTaskClientBuilder>? clientBuilder = null)
{
return services.ConfigureDurableOptions(
options => configure(options.Workflows),
workerBuilder,
clientBuilder);
}
/// <summary>
/// Configures durable agents and workflows, automatically registering orchestrations, activities, and agent entities.
/// </summary>
/// <remarks>
/// <para>
/// This is the recommended entry point for configuring durable functionality. It provides unified configuration
/// for both agents and workflows through a single <see cref="DurableOptions"/> instance, ensuring agents
/// referenced in workflows are automatically registered.
/// </para>
/// <para>
/// Multiple calls to this method (or to <see cref="ConfigureDurableAgents"/>
/// and <see cref="ConfigureDurableWorkflows"/>) are supported and configurations are composed additively.
/// </para>
/// </remarks>
/// <param name="services">The service collection to configure.</param>
/// <param name="configure">A delegate to configure the durable options for both agents and workflows.</param>
/// <param name="workerBuilder">Optional delegate to configure the durable task worker.</param>
/// <param name="clientBuilder">Optional delegate to configure the durable task client.</param>
/// <returns>The service collection for chaining.</returns>
/// <example>
/// <code>
/// services.ConfigureDurableOptions(options =>
/// {
/// // Register agents not part of workflows
/// options.Agents.AddAIAgent(standaloneAgent);
///
/// // Register workflows - agents in workflows are auto-registered
/// options.Workflows.AddWorkflow(myWorkflow);
/// },
/// workerBuilder: builder => builder.UseDurableTaskScheduler(connectionString),
/// clientBuilder: builder => builder.UseDurableTaskScheduler(connectionString));
/// </code>
/// </example>
public static IServiceCollection ConfigureDurableOptions(
this IServiceCollection services,
Action<DurableOptions> configure,
Action<IDurableTaskWorkerBuilder>? workerBuilder = null,
Action<IDurableTaskClientBuilder>? clientBuilder = null)
{
ArgumentNullException.ThrowIfNull(services);
ArgumentNullException.ThrowIfNull(configure);
DurableAgentsOptions options = services.ConfigureDurableAgents(configure);
// Get or create the shared DurableOptions instance for configuration
DurableOptions sharedOptions = GetOrCreateSharedOptions(services);
// A worker is required to run the agent entities
services.AddDurableTaskWorker(builder =>
{
workerBuilder?.Invoke(builder);
// Apply the configuration immediately to capture agent names for keyed service registration
configure(sharedOptions);
builder.AddTasks(registry =>
{
foreach (string name in options.GetAgentFactories().Keys)
{
registry.AddEntity<AgentEntity>(AgentSessionId.ToEntityName(name));
}
});
});
// Register keyed services for any new agents
RegisterAgentKeyedServices(services, sharedOptions);
// The client is needed to send notifications to the agent entities from non-orchestrator code
if (clientBuilder != null)
{
services.AddDurableTaskClient(clientBuilder);
}
services.AddSingleton<IDurableAgentClient, DefaultDurableAgentClient>();
// Register core services only once
EnsureDurableServicesRegistered(services, sharedOptions, workerBuilder, clientBuilder);
return services;
}
// This is internal because it's also used by Microsoft.Azure.Functions.DurableAgents, which is a friend assembly project.
internal static DurableAgentsOptions ConfigureDurableAgents(
this IServiceCollection services,
Action<DurableAgentsOptions> configure)
private static DurableOptions GetOrCreateSharedOptions(IServiceCollection services)
{
DurableAgentsOptions options = new();
configure(options);
// Look for an existing DurableOptions registration
ServiceDescriptor? existingDescriptor = services.FirstOrDefault(
d => d.ServiceType == typeof(DurableOptions) && d.ImplementationInstance is not null);
IReadOnlyDictionary<string, Func<IServiceProvider, AIAgent>> agents = options.GetAgentFactories();
// The agent dictionary contains the real agent factories, which is used by the agent entities.
services.AddSingleton(agents);
// Register the options so AgentEntity can access TTL configuration
services.AddSingleton(options);
// The keyed services are used to resolve durable agent *proxy* instances for external clients.
foreach (var factory in agents)
if (existingDescriptor?.ImplementationInstance is DurableOptions existing)
{
services.AddKeyedSingleton(factory.Key, (sp, _) => factory.Value(sp).AsDurableAgentProxy(sp));
return existing;
}
// A custom data converter is needed because the default chat client uses camel case for JSON properties,
// which is not the default behavior for the Durable Task SDK.
services.AddSingleton<DataConverter, DefaultDataConverter>();
// Create a new shared options instance
DurableOptions options = new();
services.AddSingleton(options);
return options;
}
private static void RegisterAgentKeyedServices(IServiceCollection services, DurableOptions options)
{
foreach (KeyValuePair<string, Func<IServiceProvider, AIAgent>> factory in options.Agents.GetAgentFactories())
{
// Only add if not already registered (to support multiple Configure* calls)
if (!services.Any(d => d.ServiceType == typeof(AIAgent) && d.IsKeyedService && Equals(d.ServiceKey, factory.Key)))
{
services.AddKeyedSingleton(factory.Key, (sp, _) => factory.Value(sp).AsDurableAgentProxy(sp));
}
}
}
/// <summary>
/// Ensures that the core durable services are registered only once, regardless of how many
/// times the configuration methods are called.
/// </summary>
private static void EnsureDurableServicesRegistered(
IServiceCollection services,
DurableOptions sharedOptions,
Action<IDurableTaskWorkerBuilder>? workerBuilder,
Action<IDurableTaskClientBuilder>? clientBuilder)
{
// Use a marker to ensure we only register core services once
if (services.Any(d => d.ServiceType == typeof(DurableServicesMarker)))
{
return;
}
services.AddSingleton<DurableServicesMarker>();
services.TryAddSingleton<DurableWorkflowRunner>();
// Configure Durable Task Worker - capture sharedOptions reference in closure.
// The options object is populated by all Configure* calls before the worker starts.
if (workerBuilder is not null)
{
services.AddDurableTaskWorker(builder =>
{
workerBuilder?.Invoke(builder);
builder.AddTasks(registry => RegisterTasksFromOptions(registry, sharedOptions));
});
}
// Configure Durable Task Client
if (clientBuilder is not null)
{
services.AddDurableTaskClient(clientBuilder);
services.TryAddSingleton<IWorkflowClient, DurableWorkflowClient>();
services.TryAddSingleton<IDurableAgentClient, DefaultDurableAgentClient>();
}
// Register workflow and agent services
services.TryAddSingleton<DataConverter, DurableDataConverter>();
// Register agent factories resolver - returns factories from the shared options
services.TryAddSingleton(
sp => sp.GetRequiredService<DurableOptions>().Agents.GetAgentFactories());
// Register DurableAgentsOptions resolver
services.TryAddSingleton(sp => sp.GetRequiredService<DurableOptions>().Agents);
}
private static void RegisterTasksFromOptions(DurableTaskRegistry registry, DurableOptions durableOptions)
{
// Build registrations for all workflows including sub-workflows
List<WorkflowRegistrationInfo> registrations = [];
HashSet<string> registeredActivities = [];
HashSet<string> registeredOrchestrations = [];
DurableWorkflowOptions workflowOptions = durableOptions.Workflows;
foreach (Workflow workflow in workflowOptions.Workflows.Values.ToList())
{
BuildWorkflowRegistrationRecursive(
workflow,
workflowOptions,
registrations,
registeredActivities,
registeredOrchestrations);
}
IReadOnlyDictionary<string, Func<IServiceProvider, AIAgent>> agentFactories =
durableOptions.Agents.GetAgentFactories();
// Register orchestrations and activities
foreach (WorkflowRegistrationInfo registration in registrations)
{
// Register with DurableWorkflowInput<object> - the DataConverter handles serialization/deserialization
registry.AddOrchestratorFunc<DurableWorkflowInput<object>, DurableWorkflowResult>(
registration.OrchestrationName,
(context, input) => RunWorkflowOrchestrationAsync(context, input, durableOptions));
foreach (ActivityRegistrationInfo activity in registration.Activities)
{
ExecutorBinding binding = activity.Binding;
registry.AddActivityFunc<string, string>(
activity.ActivityName,
(context, input) => DurableActivityExecutor.ExecuteAsync(binding, input));
}
}
// Register agent entities
foreach (string agentName in agentFactories.Keys)
{
registry.AddEntity<AgentEntity>(AgentSessionId.ToEntityName(agentName));
}
}
private static void BuildWorkflowRegistrationRecursive(
Workflow workflow,
DurableWorkflowOptions workflowOptions,
List<WorkflowRegistrationInfo> registrations,
HashSet<string> registeredActivities,
HashSet<string> registeredOrchestrations)
{
string orchestrationName = WorkflowNamingHelper.ToOrchestrationFunctionName(workflow.Name!);
if (!registeredOrchestrations.Add(orchestrationName))
{
return;
}
registrations.Add(BuildWorkflowRegistration(workflow, registeredActivities));
// Process subworkflows recursively to register them as separate orchestrations
foreach (SubworkflowBinding subworkflowBinding in workflow.ReflectExecutors()
.Select(e => e.Value)
.OfType<SubworkflowBinding>())
{
Workflow subWorkflow = subworkflowBinding.WorkflowInstance;
workflowOptions.AddWorkflow(subWorkflow);
BuildWorkflowRegistrationRecursive(
subWorkflow,
workflowOptions,
registrations,
registeredActivities,
registeredOrchestrations);
}
}
private static WorkflowRegistrationInfo BuildWorkflowRegistration(
Workflow workflow,
HashSet<string> registeredActivities)
{
string orchestrationName = WorkflowNamingHelper.ToOrchestrationFunctionName(workflow.Name!);
Dictionary<string, ExecutorBinding> executorBindings = workflow.ReflectExecutors();
List<ActivityRegistrationInfo> activities = [];
foreach (KeyValuePair<string, ExecutorBinding> entry in executorBindings
.Where(e => IsActivityBinding(e.Value)))
{
string executorName = WorkflowNamingHelper.GetExecutorName(entry.Key);
string activityName = WorkflowNamingHelper.ToOrchestrationFunctionName(executorName);
if (registeredActivities.Add(activityName))
{
activities.Add(new ActivityRegistrationInfo(activityName, entry.Value));
}
}
return new WorkflowRegistrationInfo(orchestrationName, activities);
}
/// <summary>
/// Returns <see langword="true"/> for bindings that should be registered as Durable Task activities.
/// <see cref="AIAgentBinding"/> (Durable Entities), <see cref="SubworkflowBinding"/> (sub-orchestrations),
/// and <see cref="RequestPortBinding"/> (human-in-the-loop via external events) use specialized dispatch
/// and are excluded.
/// </summary>
private static bool IsActivityBinding(ExecutorBinding binding)
=> binding is not AIAgentBinding
and not SubworkflowBinding
and not RequestPortBinding;
private static async Task<DurableWorkflowResult> RunWorkflowOrchestrationAsync(
TaskOrchestrationContext context,
DurableWorkflowInput<object> workflowInput,
DurableOptions durableOptions)
{
ILogger logger = context.CreateReplaySafeLogger("DurableWorkflow");
DurableWorkflowRunner runner = new(durableOptions);
// ConfigureAwait(true) is required in orchestration code for deterministic replay.
return await runner.RunWorkflowOrchestrationAsync(context, workflowInput, logger).ConfigureAwait(true);
}
private sealed record WorkflowRegistrationInfo(string OrchestrationName, List<ActivityRegistrationInfo> Activities);
private sealed record ActivityRegistrationInfo(string ActivityName, ExecutorBinding Binding);
/// <summary>
/// Validates that an agent with the specified name has been registered.
/// </summary>
@@ -124,63 +378,4 @@ public static class ServiceCollectionExtensions
throw new AgentNotRegisteredException(agentName);
}
}
private sealed class DefaultDataConverter : DataConverter
{
// Use durable agent options (web defaults + camel case by default) with case-insensitive matching.
// We clone to apply naming/casing tweaks while retaining source-generated metadata where available.
private static readonly JsonSerializerOptions s_options = new(DurableAgentJsonUtilities.DefaultOptions)
{
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
PropertyNameCaseInsensitive = true,
};
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Fallback path uses reflection when metadata unavailable.")]
[UnconditionalSuppressMessage("ReflectionAnalysis", "IL3050", Justification = "Fallback path uses reflection when metadata unavailable.")]
public override object? Deserialize(string? data, Type targetType)
{
if (data is null)
{
return null;
}
if (targetType == typeof(DurableAgentState))
{
return JsonSerializer.Deserialize(data, DurableAgentStateJsonContext.Default.DurableAgentState);
}
JsonTypeInfo? typeInfo = s_options.GetTypeInfo(targetType);
if (typeInfo is JsonTypeInfo typedInfo)
{
return JsonSerializer.Deserialize(data, typedInfo);
}
// Fallback (may trigger trimming/AOT warnings for unsupported dynamic types).
return JsonSerializer.Deserialize(data, targetType, s_options);
}
[return: NotNullIfNotNull(nameof(value))]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Fallback path uses reflection when metadata unavailable.")]
[UnconditionalSuppressMessage("ReflectionAnalysis", "IL3050", Justification = "Fallback path uses reflection when metadata unavailable.")]
public override string? Serialize(object? value)
{
if (value is null)
{
return null;
}
if (value is DurableAgentState durableAgentState)
{
return JsonSerializer.Serialize(durableAgentState, DurableAgentStateJsonContext.Default.DurableAgentState);
}
JsonTypeInfo? typeInfo = s_options.GetTypeInfo(value.GetType());
if (typeInfo is JsonTypeInfo typedInfo)
{
return JsonSerializer.Serialize(value, typedInfo);
}
return JsonSerializer.Serialize(value, s_options);
}
}
}
@@ -0,0 +1,177 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics.CodeAnalysis;
using System.Text.Json;
using Microsoft.Agents.AI.Workflows;
using Microsoft.Agents.AI.Workflows.Checkpointing;
using Microsoft.Agents.AI.Workflows.Observability;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Executes workflow activities by invoking executor bindings and handling serialization.
/// </summary>
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Workflow and executor types are registered at startup.")]
[UnconditionalSuppressMessage("Trimming", "IL2057", Justification = "Workflow and executor types are registered at startup.")]
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Workflow and executor types are registered at startup.")]
internal static class DurableActivityExecutor
{
/// <summary>
/// Executes an activity using the provided executor binding.
/// </summary>
/// <param name="binding">The executor binding to invoke.</param>
/// <param name="input">The serialized input string.</param>
/// <param name="cancellationToken">A token to cancel the operation.</param>
/// <returns>The serialized activity output.</returns>
/// <exception cref="ArgumentNullException">Thrown when <paramref name="binding"/> is null.</exception>
/// <exception cref="InvalidOperationException">Thrown when the executor factory is not configured.</exception>
internal static async Task<string> ExecuteAsync(
ExecutorBinding binding,
string input,
CancellationToken cancellationToken = default)
{
ArgumentNullException.ThrowIfNull(binding);
if (binding.FactoryAsync is null)
{
throw new InvalidOperationException($"Executor binding for '{binding.Id}' does not have a factory configured.");
}
DurableActivityInput? inputWithState = TryDeserializeActivityInput(input);
string executorInput = inputWithState?.Input ?? input;
Dictionary<string, string> sharedState = inputWithState?.State ?? [];
Executor executor = await binding.FactoryAsync(binding.Id).ConfigureAwait(false);
Type inputType = ResolveInputType(inputWithState?.InputTypeName, executor.InputTypes);
object typedInput = DeserializeInput(executorInput, inputType);
DurableWorkflowContext workflowContext = new(sharedState, executor);
object? result = await executor.ExecuteCoreAsync(
typedInput,
new TypeId(inputType),
workflowContext,
WorkflowTelemetryContext.Disabled,
cancellationToken).ConfigureAwait(false);
return SerializeActivityOutput(result, workflowContext);
}
private static string SerializeActivityOutput(object? result, DurableWorkflowContext context)
{
DurableExecutorOutput output = new()
{
Result = SerializeResult(result),
StateUpdates = context.StateUpdates,
ClearedScopes = [.. context.ClearedScopes],
Events = context.OutboundEvents.ConvertAll(SerializeEvent),
SentMessages = context.SentMessages,
HaltRequested = context.HaltRequested
};
return JsonSerializer.Serialize(output, DurableWorkflowJsonContext.Default.DurableExecutorOutput);
}
/// <summary>
/// Serializes a workflow event with type information for proper deserialization.
/// </summary>
private static string SerializeEvent(WorkflowEvent evt)
{
Type eventType = evt.GetType();
TypedPayload wrapper = new()
{
TypeName = eventType.AssemblyQualifiedName,
Data = JsonSerializer.Serialize(evt, eventType, DurableSerialization.Options)
};
return JsonSerializer.Serialize(wrapper, DurableWorkflowJsonContext.Default.TypedPayload);
}
private static string SerializeResult(object? result)
{
if (result is null)
{
return string.Empty;
}
if (result is string str)
{
return str;
}
return JsonSerializer.Serialize(result, result.GetType(), DurableSerialization.Options);
}
private static DurableActivityInput? TryDeserializeActivityInput(string input)
{
try
{
return JsonSerializer.Deserialize(input, DurableWorkflowJsonContext.Default.DurableActivityInput);
}
catch (JsonException)
{
return null;
}
}
internal static object DeserializeInput(string input, Type targetType)
{
if (targetType == typeof(string))
{
return input;
}
// Fan-in aggregation serializes results as a JSON array of strings (e.g., ["{...}", "{...}"]).
// When the target type is a non-string array, deserialize each element individually.
if (targetType.IsArray && targetType != typeof(string[]))
{
Type elementType = targetType.GetElementType()!;
string[]? stringArray = JsonSerializer.Deserialize<string[]>(input, DurableSerialization.Options);
if (stringArray is not null)
{
Array result = Array.CreateInstance(elementType, stringArray.Length);
for (int i = 0; i < stringArray.Length; i++)
{
object element = JsonSerializer.Deserialize(stringArray[i], elementType, DurableSerialization.Options)
?? throw new InvalidOperationException($"Failed to deserialize element {i} to type '{elementType.Name}'.");
result.SetValue(element, i);
}
return result;
}
}
return JsonSerializer.Deserialize(input, targetType, DurableSerialization.Options)
?? throw new InvalidOperationException($"Failed to deserialize input to type '{targetType.Name}'.");
}
internal static Type ResolveInputType(string? inputTypeName, ISet<Type> supportedTypes)
{
if (string.IsNullOrEmpty(inputTypeName))
{
return supportedTypes.FirstOrDefault() ?? typeof(string);
}
Type? matchedType = supportedTypes.FirstOrDefault(t =>
t.AssemblyQualifiedName == inputTypeName ||
t.FullName == inputTypeName ||
t.Name == inputTypeName);
if (matchedType is not null)
{
return matchedType;
}
Type? loadedType = Type.GetType(inputTypeName);
// Fall back if type is string or string[] but executor doesn't support it
if (loadedType is not null && !supportedTypes.Contains(loadedType))
{
if (loadedType == typeof(string) || loadedType == typeof(string[]))
{
return supportedTypes.FirstOrDefault() ?? typeof(string);
}
}
return loadedType ?? supportedTypes.FirstOrDefault() ?? typeof(string);
}
}
@@ -0,0 +1,24 @@
// Copyright (c) Microsoft. All rights reserved.
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Input payload for activity execution, containing the input and other metadata.
/// </summary>
internal sealed class DurableActivityInput
{
/// <summary>
/// Gets or sets the serialized executor input.
/// </summary>
public string? Input { get; set; }
/// <summary>
/// Gets or sets the assembly-qualified type name of the input, used for proper deserialization.
/// </summary>
public string? InputTypeName { get; set; }
/// <summary>
/// Gets or sets the shared state dictionary (scope-prefixed key -> serialized value).
/// </summary>
public Dictionary<string, string> State { get; set; } = [];
}
@@ -0,0 +1,216 @@
// Copyright (c) Microsoft. All rights reserved.
// ConfigureAwait Usage in Orchestration Code:
// This file uses ConfigureAwait(true) because it runs within orchestration context.
// Durable Task orchestrations require deterministic replay - the same code must execute
// identically across replays. ConfigureAwait(true) ensures continuations run on the
// orchestration's synchronization context, which is essential for replay correctness.
// Using ConfigureAwait(false) here could cause non-deterministic behavior during replay.
using System.Text.Json;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask;
using Microsoft.Extensions.Logging;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Dispatches workflow executors to activities, AI agents, sub-orchestrations, or external events (human-in-the-loop).
/// </summary>
/// <remarks>
/// Called during the dispatch phase of each superstep by
/// <c>DurableWorkflowRunner.DispatchExecutorsInParallelAsync</c>. For each executor that has
/// pending input, this dispatcher determines whether the executor is an AI agent (stateful,
/// backed by Durable Entities), a request port (human-in-the-loop, backed by external events),
/// a sub-workflow (dispatched as a sub-orchestration), or a regular activity, and invokes the
/// appropriate Durable Task API.
/// The serialised string result is returned to the runner for the routing phase.
/// </remarks>
internal static class DurableExecutorDispatcher
{
/// <summary>
/// Dispatches an executor based on its type (activity, AI agent, request port, or sub-workflow).
/// </summary>
/// <param name="context">The task orchestration context.</param>
/// <param name="executorInfo">Information about the executor to dispatch.</param>
/// <param name="envelope">The message envelope containing input and type information.</param>
/// <param name="sharedState">The shared state dictionary to pass to the executor.</param>
/// <param name="liveStatus">The live workflow status used to publish events and pending request port state.</param>
/// <param name="logger">The logger for tracing.</param>
/// <returns>The result from the executor.</returns>
internal static async Task<string> DispatchAsync(
TaskOrchestrationContext context,
WorkflowExecutorInfo executorInfo,
DurableMessageEnvelope envelope,
Dictionary<string, string> sharedState,
DurableWorkflowLiveStatus liveStatus,
ILogger logger)
{
logger.LogDispatchingExecutor(executorInfo.ExecutorId, executorInfo.IsAgenticExecutor);
if (executorInfo.IsRequestPortExecutor)
{
return await ExecuteRequestPortAsync(context, executorInfo, envelope.Message, liveStatus, logger).ConfigureAwait(true);
}
if (executorInfo.IsAgenticExecutor)
{
return await ExecuteAgentAsync(context, executorInfo, logger, envelope.Message).ConfigureAwait(true);
}
if (executorInfo.IsSubworkflowExecutor)
{
return await ExecuteSubWorkflowAsync(context, executorInfo, envelope.Message).ConfigureAwait(true);
}
return await ExecuteActivityAsync(context, executorInfo, envelope.Message, envelope.InputTypeName, sharedState).ConfigureAwait(true);
}
private static async Task<string> ExecuteActivityAsync(
TaskOrchestrationContext context,
WorkflowExecutorInfo executorInfo,
string input,
string? inputTypeName,
Dictionary<string, string> sharedState)
{
string executorName = WorkflowNamingHelper.GetExecutorName(executorInfo.ExecutorId);
string activityName = WorkflowNamingHelper.ToOrchestrationFunctionName(executorName);
DurableActivityInput activityInput = new()
{
Input = input,
InputTypeName = inputTypeName,
State = sharedState
};
string serializedInput = JsonSerializer.Serialize(activityInput, DurableWorkflowJsonContext.Default.DurableActivityInput);
return await context.CallActivityAsync<string>(activityName, serializedInput).ConfigureAwait(true);
}
/// <summary>
/// Executes a request port executor by waiting for an external event (human-in-the-loop).
/// </summary>
/// <remarks>
/// When the workflow reaches a <see cref="RequestPort"/> executor, the orchestration publishes
/// the pending request to <see cref="DurableWorkflowLiveStatus"/> and waits for an external actor
/// (e.g., a UI or API) to raise the corresponding event via
/// <see cref="IStreamingWorkflowRun.SendResponseAsync{TResponse}(DurableWorkflowWaitingForInputEvent, TResponse, CancellationToken)"/>.
/// Multiple RequestPorts may be dispatched in parallel during a fan-out superstep.
/// Each adds its pending request to <see cref="DurableWorkflowLiveStatus.PendingEvents"/>.
/// The wait has no built-in timeout; for time-limited approvals, callers can combine
/// <c>context.CreateTimer</c> with <c>Task.WhenAny</c> in a wrapper executor.
/// </remarks>
private static async Task<string> ExecuteRequestPortAsync(
TaskOrchestrationContext context,
WorkflowExecutorInfo executorInfo,
string input,
DurableWorkflowLiveStatus liveStatus,
ILogger logger)
{
RequestPort requestPort = executorInfo.RequestPort!;
string eventName = requestPort.Id;
logger.LogWaitingForExternalEvent(eventName);
// Publish pending request so external clients can discover what input is needed
liveStatus.PendingEvents.Add(new PendingRequestPortStatus(EventName: eventName, Input: input));
context.SetCustomStatus(liveStatus);
// Wait until the external actor raises the event
string response = await context.WaitForExternalEvent<string>(eventName).ConfigureAwait(true);
// Remove this pending request after receiving the response
liveStatus.PendingEvents.RemoveAll(p => p.EventName == eventName);
context.SetCustomStatus(liveStatus.Events.Count > 0 || liveStatus.PendingEvents.Count > 0 ? liveStatus : null);
logger.LogReceivedExternalEvent(eventName);
return response;
}
/// <summary>
/// Executes an AI agent executor through Durable Entities.
/// </summary>
/// <remarks>
/// AI agents are stateful and maintain conversation history. They use Durable Entities
/// to persist state across orchestration replays.
/// </remarks>
private static async Task<string> ExecuteAgentAsync(
TaskOrchestrationContext context,
WorkflowExecutorInfo executorInfo,
ILogger logger,
string input)
{
string agentName = WorkflowNamingHelper.GetExecutorName(executorInfo.ExecutorId);
DurableAIAgent agent = context.GetAgent(agentName);
if (agent is null)
{
logger.LogAgentNotFound(agentName);
return $"Agent '{agentName}' not found";
}
AgentSession session = await agent.CreateSessionAsync().ConfigureAwait(true);
AgentResponse response = await agent.RunAsync(input, session).ConfigureAwait(true);
return response.Text;
}
/// <summary>
/// Dispatches a sub-workflow executor as a sub-orchestration.
/// </summary>
/// <remarks>
/// Sub-workflows run as separate orchestration instances, providing independent
/// checkpointing, replay, and hierarchical visualization in the DTS dashboard.
/// The input is wrapped in <see cref="DurableWorkflowInput{T}"/> so the sub-orchestration
/// can extract it using the same envelope structure. The sub-orchestration returns a
/// <see cref="DurableWorkflowResult"/> directly (deserialized by the Durable Task SDK),
/// which this method converts to a <see cref="DurableExecutorOutput"/> so the parent
/// workflow's result processing picks up both the result and any accumulated events.
/// </remarks>
private static async Task<string> ExecuteSubWorkflowAsync(
TaskOrchestrationContext context,
WorkflowExecutorInfo executorInfo,
string input)
{
string orchestrationName = WorkflowNamingHelper.ToOrchestrationFunctionName(executorInfo.SubWorkflow!.Name!);
DurableWorkflowInput<string> workflowInput = new() { Input = input };
DurableWorkflowResult? workflowResult = await context.CallSubOrchestratorAsync<DurableWorkflowResult?>(
orchestrationName,
workflowInput).ConfigureAwait(true);
return ConvertWorkflowResultToExecutorOutput(workflowResult);
}
/// <summary>
/// Converts a <see cref="DurableWorkflowResult"/> from a sub-orchestration
/// into a <see cref="DurableExecutorOutput"/> JSON string. This bridges the sub-workflow's
/// output format to the parent workflow's result processing, preserving both the result
/// and any accumulated events from the sub-workflow.
/// </summary>
private static string ConvertWorkflowResultToExecutorOutput(DurableWorkflowResult? workflowResult)
{
if (workflowResult is null)
{
return string.Empty;
}
// Propagate the result, events, and sent messages from the sub-workflow.
// SentMessages carry the sub-workflow's output for typed routing in the parent,
// matching the in-process WorkflowHostExecutor behavior.
// Shared state is not included because each workflow instance maintains its own
// independent shared state; it is not shared between parent and sub-workflows.
DurableExecutorOutput executorOutput = new()
{
Result = workflowResult.Result,
Events = workflowResult.Events ?? [],
SentMessages = workflowResult.SentMessages ?? [],
HaltRequested = workflowResult.HaltRequested,
};
return JsonSerializer.Serialize(executorOutput, DurableWorkflowJsonContext.Default.DurableExecutorOutput);
}
}
@@ -0,0 +1,39 @@
// Copyright (c) Microsoft. All rights reserved.
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Output payload from executor execution, containing the result, state updates, and emitted events.
/// </summary>
internal sealed class DurableExecutorOutput
{
/// <summary>
/// Gets the executor result.
/// </summary>
public string? Result { get; init; }
/// <summary>
/// Gets the state updates (scope-prefixed key to value; null indicates deletion).
/// </summary>
public Dictionary<string, string?> StateUpdates { get; init; } = [];
/// <summary>
/// Gets the scope names that were cleared.
/// </summary>
public List<string> ClearedScopes { get; init; } = [];
/// <summary>
/// Gets the workflow events emitted during execution.
/// </summary>
public List<string> Events { get; init; } = [];
/// <summary>
/// Gets the typed messages sent to downstream executors.
/// </summary>
public List<TypedPayload> SentMessages { get; init; } = [];
/// <summary>
/// Gets a value indicating whether the executor requested a workflow halt.
/// </summary>
public bool HaltRequested { get; init; }
}
@@ -0,0 +1,25 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Event raised when an executor requests the workflow to halt via <see cref="IWorkflowContext.RequestHaltAsync"/>.
/// </summary>
public sealed class DurableHaltRequestedEvent : WorkflowEvent
{
/// <summary>
/// Initializes a new instance of the <see cref="DurableHaltRequestedEvent"/> class.
/// </summary>
/// <param name="executorId">The ID of the executor that requested the halt.</param>
public DurableHaltRequestedEvent(string executorId) : base($"Halt requested by {executorId}")
{
this.ExecutorId = executorId;
}
/// <summary>
/// Gets the ID of the executor that requested the halt.
/// </summary>
public string ExecutorId { get; }
}
@@ -0,0 +1,51 @@
// Copyright (c) Microsoft. All rights reserved.
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Represents a message envelope for durable workflow message passing.
/// </summary>
/// <remarks>
/// <para>
/// This is the durable equivalent of <c>MessageEnvelope</c> in the in-process runner.
/// Unlike the in-process version which holds native .NET objects, this envelope
/// contains serialized JSON strings suitable for Durable Task activities.
/// </para>
/// </remarks>
internal sealed class DurableMessageEnvelope
{
/// <summary>
/// Gets or sets the serialized JSON message content.
/// </summary>
public required string Message { get; init; }
/// <summary>
/// Gets or sets the full type name of the message for deserialization.
/// </summary>
public string? InputTypeName { get; init; }
/// <summary>
/// Gets or sets the ID of the executor that produced this message.
/// </summary>
/// <remarks>
/// Used for tracing and debugging. Null for initial workflow input.
/// </remarks>
public string? SourceExecutorId { get; init; }
/// <summary>
/// Creates a new message envelope.
/// </summary>
/// <param name="message">The serialized JSON message content.</param>
/// <param name="inputTypeName">The full type name of the message for deserialization.</param>
/// <param name="sourceExecutorId">The ID of the executor that produced this message, or null for initial input.</param>
/// <returns>A new <see cref="DurableMessageEnvelope"/> instance.</returns>
internal static DurableMessageEnvelope Create(string message, string? inputTypeName, string? sourceExecutorId = null)
{
return new DurableMessageEnvelope
{
Message = message,
InputTypeName = inputTypeName,
SourceExecutorId = sourceExecutorId
};
}
}
@@ -0,0 +1,49 @@
// Copyright (c) Microsoft. All rights reserved.
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Represents the execution status of a durable workflow run.
/// </summary>
public enum DurableRunStatus
{
/// <summary>
/// The workflow instance was not found.
/// </summary>
NotFound,
/// <summary>
/// The workflow is pending and has not started.
/// </summary>
Pending,
/// <summary>
/// The workflow is currently running.
/// </summary>
Running,
/// <summary>
/// The workflow completed successfully.
/// </summary>
Completed,
/// <summary>
/// The workflow failed with an error.
/// </summary>
Failed,
/// <summary>
/// The workflow was terminated.
/// </summary>
Terminated,
/// <summary>
/// The workflow is suspended.
/// </summary>
Suspended,
/// <summary>
/// The workflow status is unknown.
/// </summary>
Unknown
}
@@ -0,0 +1,22 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Text.Json;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Shared serialization options for user-defined workflow types that are not known at compile time
/// and therefore cannot use the source-generated <see cref="DurableWorkflowJsonContext"/>.
/// </summary>
internal static class DurableSerialization
{
/// <summary>
/// Gets the shared <see cref="JsonSerializerOptions"/> for workflow serialization
/// with camelCase naming and case-insensitive deserialization.
/// </summary>
internal static JsonSerializerOptions Options { get; } = new()
{
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
PropertyNameCaseInsensitive = true
};
}
@@ -0,0 +1,452 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics;
using System.Diagnostics.CodeAnalysis;
using System.Runtime.CompilerServices;
using System.Text.Json;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask;
using Microsoft.DurableTask.Client;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Represents a durable workflow run that supports streaming workflow events as they occur.
/// </summary>
/// <remarks>
/// <para>
/// Events are detected by monitoring the orchestration's custom status at regular intervals.
/// When executors emit events via <see cref="IWorkflowContext.AddEventAsync"/> or
/// <see cref="IWorkflowContext.YieldOutputAsync"/>, they are written to the orchestration's
/// custom status and picked up by this streaming run.
/// </para>
/// <para>
/// When the workflow reaches a <see cref="RequestPort"/> executor, a <see cref="DurableWorkflowWaitingForInputEvent"/>
/// is yielded containing the request data. The caller should then call
/// <see cref="SendResponseAsync{TResponse}(DurableWorkflowWaitingForInputEvent, TResponse, CancellationToken)"/>
/// to provide the response and resume the workflow.
/// </para>
/// </remarks>
[DebuggerDisplay("{WorkflowName} ({RunId})")]
internal sealed class DurableStreamingWorkflowRun : IStreamingWorkflowRun
{
private readonly DurableTaskClient _client;
private readonly Dictionary<string, RequestPort> _requestPorts;
/// <summary>
/// Initializes a new instance of the <see cref="DurableStreamingWorkflowRun"/> class.
/// </summary>
/// <param name="client">The durable task client for orchestration operations.</param>
/// <param name="instanceId">The unique instance ID for this orchestration run.</param>
/// <param name="workflow">The workflow being executed.</param>
internal DurableStreamingWorkflowRun(DurableTaskClient client, string instanceId, Workflow workflow)
{
this._client = client;
this.RunId = instanceId;
this.WorkflowName = workflow.Name ?? string.Empty;
this._requestPorts = ExtractRequestPorts(workflow);
}
/// <inheritdoc/>
public string RunId { get; }
/// <summary>
/// Gets the name of the workflow being executed.
/// </summary>
public string WorkflowName { get; }
/// <summary>
/// Gets the current execution status of the workflow run.
/// </summary>
/// <param name="cancellationToken">A cancellation token to observe.</param>
/// <returns>The current status of the durable run.</returns>
public async ValueTask<DurableRunStatus> GetStatusAsync(CancellationToken cancellationToken = default)
{
OrchestrationMetadata? metadata = await this._client.GetInstanceAsync(
this.RunId,
getInputsAndOutputs: false,
cancellation: cancellationToken).ConfigureAwait(false);
if (metadata is null)
{
return DurableRunStatus.NotFound;
}
return metadata.RuntimeStatus switch
{
OrchestrationRuntimeStatus.Pending => DurableRunStatus.Pending,
OrchestrationRuntimeStatus.Running => DurableRunStatus.Running,
OrchestrationRuntimeStatus.Completed => DurableRunStatus.Completed,
OrchestrationRuntimeStatus.Failed => DurableRunStatus.Failed,
OrchestrationRuntimeStatus.Terminated => DurableRunStatus.Terminated,
OrchestrationRuntimeStatus.Suspended => DurableRunStatus.Suspended,
_ => DurableRunStatus.Unknown
};
}
/// <inheritdoc/>
public IAsyncEnumerable<WorkflowEvent> WatchStreamAsync(CancellationToken cancellationToken = default)
=> this.WatchStreamAsync(pollingInterval: null, cancellationToken);
/// <summary>
/// Asynchronously streams workflow events as they occur during workflow execution.
/// </summary>
/// <param name="pollingInterval">The interval between status checks. Defaults to 100ms.</param>
/// <param name="cancellationToken">A cancellation token to observe.</param>
/// <returns>An asynchronous stream of <see cref="WorkflowEvent"/> objects.</returns>
private async IAsyncEnumerable<WorkflowEvent> WatchStreamAsync(
TimeSpan? pollingInterval,
[EnumeratorCancellation] CancellationToken cancellationToken = default)
{
TimeSpan minInterval = pollingInterval ?? TimeSpan.FromMilliseconds(100);
TimeSpan maxInterval = TimeSpan.FromSeconds(2);
TimeSpan currentInterval = minInterval;
// Track how many events we've already read from the durable workflow status
int lastReadEventIndex = 0;
// Track which pending events we've already yielded to avoid duplicates
HashSet<string> yieldedPendingEvents = [];
while (!cancellationToken.IsCancellationRequested)
{
// Poll with getInputsAndOutputs: true because SerializedCustomStatus
// (used for event streaming) is only populated when this flag is set.
OrchestrationMetadata? metadata = await this._client.GetInstanceAsync(
this.RunId,
getInputsAndOutputs: true,
cancellation: cancellationToken).ConfigureAwait(false);
if (metadata is null)
{
yield break;
}
bool hasNewEvents = false;
// Always drain any unread events from the durable workflow status before checking terminal states.
// The orchestration may complete before the next poll, so events would be lost if we
// check terminal status first.
if (metadata.SerializedCustomStatus is not null)
{
if (DurableWorkflowLiveStatus.TryParse(metadata.SerializedCustomStatus, out DurableWorkflowLiveStatus liveStatus))
{
(List<WorkflowEvent> events, lastReadEventIndex) = DrainNewEvents(liveStatus.Events, lastReadEventIndex);
foreach (WorkflowEvent evt in events)
{
hasNewEvents = true;
yield return evt;
}
// Yield a DurableWorkflowWaitingForInputEvent for each new pending request port
foreach (PendingRequestPortStatus pending in liveStatus.PendingEvents)
{
if (yieldedPendingEvents.Add(pending.EventName))
{
if (!this._requestPorts.TryGetValue(pending.EventName, out RequestPort? matchingPort))
{
// RequestPort may not exist in the current workflow definition (e.g., during rolling deployments).
continue;
}
hasNewEvents = true;
yield return new DurableWorkflowWaitingForInputEvent(
pending.Input,
matchingPort);
}
}
// Sync tracking with current pending events so re-used RequestPort names can be yielded again
if (liveStatus.PendingEvents.Count == 0)
{
yieldedPendingEvents.Clear();
}
else
{
yieldedPendingEvents.IntersectWith(liveStatus.PendingEvents.Select(p => p.EventName));
}
}
}
// Check terminal states after draining events from the durable workflow status
if (metadata.RuntimeStatus == OrchestrationRuntimeStatus.Completed)
{
// The framework clears the durable workflow status on completion, so events may be in
// SerializedOutput as a DurableWorkflowResult wrapper.
if (TryParseWorkflowResult(metadata.SerializedOutput, out DurableWorkflowResult? outputResult))
{
(List<WorkflowEvent> events, _) = DrainNewEvents(outputResult.Events, lastReadEventIndex);
foreach (WorkflowEvent evt in events)
{
yield return evt;
}
yield return new DurableWorkflowCompletedEvent(outputResult.Result);
}
else
{
// The runner always wraps output in DurableWorkflowResult, so a parse
// failure here indicates a bug. Yield a failed event so the consumer
// gets a visible, handleable signal without crashing.
yield return new DurableWorkflowFailedEvent(
$"Workflow '{this.WorkflowName}' (RunId: {this.RunId}) completed but its output could not be parsed as DurableWorkflowResult.");
}
yield break;
}
if (metadata.RuntimeStatus == OrchestrationRuntimeStatus.Failed)
{
string errorMessage = metadata.FailureDetails?.ErrorMessage ?? "Workflow execution failed.";
yield return new DurableWorkflowFailedEvent(errorMessage, metadata.FailureDetails);
yield break;
}
if (metadata.RuntimeStatus == OrchestrationRuntimeStatus.Terminated)
{
yield return new DurableWorkflowFailedEvent("Workflow was terminated.");
yield break;
}
// Adaptive backoff: reset to minimum when events were found, increase otherwise
currentInterval = hasNewEvents
? minInterval
: TimeSpan.FromMilliseconds(Math.Min(currentInterval.TotalMilliseconds * 2, maxInterval.TotalMilliseconds));
try
{
await Task.Delay(currentInterval, cancellationToken).ConfigureAwait(false);
}
catch (OperationCanceledException) when (cancellationToken.IsCancellationRequested)
{
yield break;
}
}
}
/// <summary>
/// Sends a response to a <see cref="DurableWorkflowWaitingForInputEvent"/> to resume the workflow.
/// </summary>
/// <typeparam name="TResponse">The type of the response data.</typeparam>
/// <param name="requestEvent">The request event to respond to.</param>
/// <param name="response">The response data to send.</param>
/// <param name="cancellationToken">A cancellation token to observe.</param>
/// <returns>A <see cref="ValueTask"/> representing the asynchronous operation.</returns>
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Serializing workflow types provided by the caller.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Serializing workflow types provided by the caller.")]
public async ValueTask SendResponseAsync<TResponse>(DurableWorkflowWaitingForInputEvent requestEvent, TResponse response, CancellationToken cancellationToken = default)
{
ArgumentNullException.ThrowIfNull(requestEvent);
string serializedResponse = JsonSerializer.Serialize(response, DurableSerialization.Options);
await this._client.RaiseEventAsync(
this.RunId,
requestEvent.RequestPort.Id,
serializedResponse,
cancellationToken).ConfigureAwait(false);
}
/// <summary>
/// Waits for the workflow to complete and returns the result.
/// </summary>
/// <typeparam name="TResult">The expected result type.</typeparam>
/// <param name="cancellationToken">A cancellation token to observe.</param>
/// <returns>The result of the workflow execution.</returns>
/// <exception cref="TaskFailedException">Thrown when the workflow failed.</exception>
/// <exception cref="InvalidOperationException">Thrown when the workflow was terminated or ended with an unexpected status.</exception>
public async ValueTask<TResult?> WaitForCompletionAsync<TResult>(CancellationToken cancellationToken = default)
{
OrchestrationMetadata metadata = await this._client.WaitForInstanceCompletionAsync(
this.RunId,
getInputsAndOutputs: true,
cancellation: cancellationToken).ConfigureAwait(false);
if (metadata.RuntimeStatus == OrchestrationRuntimeStatus.Completed)
{
return ExtractResult<TResult>(metadata.SerializedOutput);
}
if (metadata.RuntimeStatus == OrchestrationRuntimeStatus.Failed)
{
if (metadata.FailureDetails is not null)
{
throw new TaskFailedException(
taskName: this.WorkflowName,
taskId: -1,
failureDetails: metadata.FailureDetails);
}
throw new InvalidOperationException(
$"Workflow '{this.WorkflowName}' (RunId: {this.RunId}) failed without failure details.");
}
throw new InvalidOperationException(
$"Workflow '{this.WorkflowName}' (RunId: {this.RunId}) ended with unexpected status: {metadata.RuntimeStatus}");
}
/// <summary>
/// Deserializes and returns any events beyond <paramref name="lastReadIndex"/> from the list.
/// </summary>
private static (List<WorkflowEvent> Events, int UpdatedIndex) DrainNewEvents(List<string> serializedEvents, int lastReadIndex)
{
List<WorkflowEvent> events = [];
while (lastReadIndex < serializedEvents.Count)
{
string serializedEvent = serializedEvents[lastReadIndex];
lastReadIndex++;
WorkflowEvent? workflowEvent = TryDeserializeEvent(serializedEvent);
if (workflowEvent is not null)
{
events.Add(workflowEvent);
}
}
return (events, lastReadIndex);
}
/// <summary>
/// Attempts to parse the orchestration output as a <see cref="DurableWorkflowResult"/> wrapper.
/// </summary>
/// <remarks>
/// The orchestration returns a <see cref="DurableWorkflowResult"/> object directly.
/// The Durable Task framework's <c>DataConverter</c> serializes it as a JSON object
/// in <c>SerializedOutput</c>, so we deserialize it directly.
/// </remarks>
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Deserializing workflow result wrapper.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Deserializing workflow result wrapper.")]
private static bool TryParseWorkflowResult(string? serializedOutput, [NotNullWhen(true)] out DurableWorkflowResult? result)
{
if (serializedOutput is null)
{
result = default!;
return false;
}
try
{
result = JsonSerializer.Deserialize(serializedOutput, DurableWorkflowJsonContext.Default.DurableWorkflowResult)!;
return result is not null;
}
catch (JsonException)
{
result = default!;
return false;
}
}
/// <summary>
/// Extracts a typed result from the orchestration output by unwrapping the
/// <see cref="DurableWorkflowResult"/> wrapper.
/// </summary>
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Deserializing workflow result.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Deserializing workflow result.")]
internal static TResult? ExtractResult<TResult>(string? serializedOutput)
{
if (serializedOutput is null)
{
return default;
}
if (!TryParseWorkflowResult(serializedOutput, out DurableWorkflowResult? workflowResult))
{
throw new InvalidOperationException(
"Failed to parse orchestration output as DurableWorkflowResult. " +
"The orchestration runner should always wrap output in this format.");
}
string? resultJson = workflowResult.Result;
if (resultJson is null)
{
return default;
}
if (typeof(TResult) == typeof(string))
{
return (TResult)(object)resultJson;
}
return JsonSerializer.Deserialize<TResult>(resultJson, DurableSerialization.Options);
}
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Deserializing workflow event types.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Deserializing workflow event types.")]
[UnconditionalSuppressMessage("Trimming", "IL2057", Justification = "Event types are registered at startup.")]
private static WorkflowEvent? TryDeserializeEvent(string serializedEvent)
{
try
{
TypedPayload? wrapper = JsonSerializer.Deserialize(
serializedEvent,
DurableWorkflowJsonContext.Default.TypedPayload);
if (wrapper?.TypeName is not null && wrapper.Data is not null)
{
Type? eventType = Type.GetType(wrapper.TypeName);
if (eventType is not null)
{
return DeserializeEventByType(eventType, wrapper.Data);
}
}
return null;
}
catch (JsonException)
{
return null;
}
}
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Deserializing workflow event types.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Deserializing workflow event types.")]
private static WorkflowEvent? DeserializeEventByType(Type eventType, string json)
{
// Types with internal constructors need manual deserialization
if (eventType == typeof(ExecutorInvokedEvent)
|| eventType == typeof(ExecutorCompletedEvent)
|| eventType == typeof(WorkflowOutputEvent))
{
using JsonDocument doc = JsonDocument.Parse(json);
JsonElement root = doc.RootElement;
if (eventType == typeof(ExecutorInvokedEvent))
{
string executorId = root.GetProperty("executorId").GetString() ?? string.Empty;
JsonElement? data = GetDataProperty(root);
return new ExecutorInvokedEvent(executorId, data!);
}
if (eventType == typeof(ExecutorCompletedEvent))
{
string executorId = root.GetProperty("executorId").GetString() ?? string.Empty;
JsonElement? data = GetDataProperty(root);
return new ExecutorCompletedEvent(executorId, data);
}
// WorkflowOutputEvent
string sourceId = root.GetProperty("sourceId").GetString() ?? string.Empty;
object? outputData = GetDataProperty(root);
return new WorkflowOutputEvent(outputData!, sourceId);
}
return JsonSerializer.Deserialize(json, eventType, DurableSerialization.Options) as WorkflowEvent;
}
private static JsonElement? GetDataProperty(JsonElement root)
{
if (!root.TryGetProperty("data", out JsonElement dataElement))
{
return null;
}
return dataElement.ValueKind == JsonValueKind.Null ? null : dataElement.Clone();
}
private static Dictionary<string, RequestPort> ExtractRequestPorts(Workflow workflow)
{
return WorkflowAnalyzer.GetExecutorsFromWorkflowInOrder(workflow)
.Where(e => e.RequestPort is not null)
.ToDictionary(e => e.RequestPort!.Id, e => e.RequestPort!);
}
}
@@ -0,0 +1,95 @@
// Copyright (c) Microsoft. All rights reserved.
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask;
using Microsoft.DurableTask.Client;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Provides a durable task-based implementation of <see cref="IWorkflowClient"/> for running
/// workflows as durable orchestrations.
/// </summary>
internal sealed class DurableWorkflowClient : IWorkflowClient
{
private readonly DurableTaskClient _client;
/// <summary>
/// Initializes a new instance of the <see cref="DurableWorkflowClient"/> class.
/// </summary>
/// <param name="client">The durable task client for orchestration operations.</param>
/// <exception cref="ArgumentNullException">Thrown when <paramref name="client"/> is null.</exception>
public DurableWorkflowClient(DurableTaskClient client)
{
ArgumentNullException.ThrowIfNull(client);
this._client = client;
}
/// <inheritdoc/>
public async ValueTask<IWorkflowRun> RunAsync<TInput>(
Workflow workflow,
TInput input,
string? runId = null,
CancellationToken cancellationToken = default)
where TInput : notnull
{
ArgumentNullException.ThrowIfNull(workflow);
if (string.IsNullOrEmpty(workflow.Name))
{
throw new ArgumentException("Workflow must have a valid Name property.", nameof(workflow));
}
DurableWorkflowInput<TInput> workflowInput = new() { Input = input };
string instanceId = await this._client.ScheduleNewOrchestrationInstanceAsync(
orchestratorName: WorkflowNamingHelper.ToOrchestrationFunctionName(workflow.Name),
input: workflowInput,
options: runId is not null ? new StartOrchestrationOptions(runId) : null,
cancellation: cancellationToken).ConfigureAwait(false);
return new DurableWorkflowRun(this._client, instanceId, workflow.Name);
}
/// <inheritdoc/>
public ValueTask<IWorkflowRun> RunAsync(
Workflow workflow,
string input,
string? runId = null,
CancellationToken cancellationToken = default)
=> this.RunAsync<string>(workflow, input, runId, cancellationToken);
/// <inheritdoc/>
public async ValueTask<IStreamingWorkflowRun> StreamAsync<TInput>(
Workflow workflow,
TInput input,
string? runId = null,
CancellationToken cancellationToken = default)
where TInput : notnull
{
ArgumentNullException.ThrowIfNull(workflow);
if (string.IsNullOrEmpty(workflow.Name))
{
throw new ArgumentException("Workflow must have a valid Name property.", nameof(workflow));
}
DurableWorkflowInput<TInput> workflowInput = new() { Input = input };
string instanceId = await this._client.ScheduleNewOrchestrationInstanceAsync(
orchestratorName: WorkflowNamingHelper.ToOrchestrationFunctionName(workflow.Name),
input: workflowInput,
options: runId is not null ? new StartOrchestrationOptions(runId) : null,
cancellation: cancellationToken).ConfigureAwait(false);
return new DurableStreamingWorkflowRun(this._client, instanceId, workflow);
}
/// <inheritdoc/>
public ValueTask<IStreamingWorkflowRun> StreamAsync(
Workflow workflow,
string input,
string? runId = null,
CancellationToken cancellationToken = default)
=> this.StreamAsync<string>(workflow, input, runId, cancellationToken);
}
@@ -0,0 +1,27 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics;
using Microsoft.Agents.AI.Workflows;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Event raised when a durable workflow completes successfully.
/// </summary>
[DebuggerDisplay("Completed: {Result}")]
public sealed class DurableWorkflowCompletedEvent : WorkflowEvent
{
/// <summary>
/// Initializes a new instance of the <see cref="DurableWorkflowCompletedEvent"/> class.
/// </summary>
/// <param name="result">The serialized result of the workflow.</param>
public DurableWorkflowCompletedEvent(string? result) : base(result)
{
this.Result = result;
}
/// <summary>
/// Gets the serialized result of the workflow.
/// </summary>
public string? Result { get; }
}
@@ -0,0 +1,327 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics;
using System.Diagnostics.CodeAnalysis;
using System.Text.Json;
using Microsoft.Agents.AI.Workflows;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// A workflow context for durable workflow execution.
/// </summary>
/// <remarks>
/// State is passed in from the orchestration and updates are collected for return.
/// Events emitted during execution are collected and returned to the orchestration
/// as part of the activity output for streaming to callers.
/// </remarks>
[DebuggerDisplay("Executor = {_executor.Id}, StateEntries = {_initialState.Count}")]
internal sealed class DurableWorkflowContext : IWorkflowContext
{
/// <summary>
/// The default scope name used when no explicit scope is specified.
/// Scopes partition shared state into logical namespaces so that different
/// parts of a workflow can manage their state keys independently.
/// </summary>
private const string DefaultScopeName = "__default__";
private readonly Dictionary<string, string> _initialState;
private readonly Executor _executor;
/// <summary>
/// Initializes a new instance of the <see cref="DurableWorkflowContext"/> class.
/// </summary>
/// <param name="initialState">The shared state passed from the orchestration.</param>
/// <param name="executor">The executor running in this context.</param>
internal DurableWorkflowContext(Dictionary<string, string>? initialState, Executor executor)
{
this._executor = executor;
this._initialState = initialState ?? [];
}
/// <summary>
/// Gets the messages sent during activity execution via <see cref="SendMessageAsync"/>.
/// </summary>
internal List<TypedPayload> SentMessages { get; } = [];
/// <summary>
/// Gets the outbound events that were added during activity execution.
/// </summary>
internal List<WorkflowEvent> OutboundEvents { get; } = [];
/// <summary>
/// Gets the state updates made during activity execution.
/// </summary>
internal Dictionary<string, string?> StateUpdates { get; } = [];
/// <summary>
/// Gets the scopes that were cleared during activity execution.
/// </summary>
internal HashSet<string> ClearedScopes { get; } = [];
/// <summary>
/// Gets a value indicating whether the executor requested a workflow halt.
/// </summary>
internal bool HaltRequested { get; private set; }
/// <inheritdoc/>
public ValueTask AddEventAsync(
WorkflowEvent workflowEvent,
CancellationToken cancellationToken = default)
{
if (workflowEvent is not null)
{
this.OutboundEvents.Add(workflowEvent);
}
return default;
}
/// <inheritdoc/>
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Serializing workflow message types registered at startup.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Serializing workflow message types registered at startup.")]
public ValueTask SendMessageAsync(
object message,
string? targetId = null,
CancellationToken cancellationToken = default)
{
if (message is not null)
{
Type messageType = message.GetType();
this.SentMessages.Add(new TypedPayload
{
Data = JsonSerializer.Serialize(message, messageType, DurableSerialization.Options),
TypeName = messageType.AssemblyQualifiedName
});
}
return default;
}
/// <inheritdoc/>
public ValueTask YieldOutputAsync(
object output,
CancellationToken cancellationToken = default)
{
if (output is not null)
{
Type outputType = output.GetType();
if (!this._executor.CanOutput(outputType))
{
throw new InvalidOperationException(
$"Cannot output object of type {outputType.Name}. " +
$"Expecting one of [{string.Join(", ", this._executor.OutputTypes)}].");
}
this.OutboundEvents.Add(new WorkflowOutputEvent(output, this._executor.Id));
}
return default;
}
/// <inheritdoc/>
public ValueTask RequestHaltAsync()
{
this.HaltRequested = true;
this.OutboundEvents.Add(new DurableHaltRequestedEvent(this._executor.Id));
return default;
}
/// <inheritdoc/>
public ValueTask<T?> ReadStateAsync<T>(
string key,
string? scopeName = null,
CancellationToken cancellationToken = default)
{
ArgumentException.ThrowIfNullOrEmpty(key);
string scopeKey = GetScopeKey(scopeName, key);
string normalizedScope = scopeName ?? DefaultScopeName;
bool scopeCleared = this.ClearedScopes.Contains(normalizedScope);
// Local updates take priority over initial state.
if (this.StateUpdates.TryGetValue(scopeKey, out string? updated))
{
return DeserializeStateAsync<T>(updated);
}
// If scope was cleared, ignore initial state
if (scopeCleared)
{
return ValueTask.FromResult<T?>(default);
}
// Fall back to initial state passed from orchestration
if (this._initialState.TryGetValue(scopeKey, out string? initial))
{
return DeserializeStateAsync<T>(initial);
}
return ValueTask.FromResult<T?>(default);
}
/// <inheritdoc/>
public async ValueTask<T> ReadOrInitStateAsync<T>(
string key,
Func<T> initialStateFactory,
string? scopeName = null,
CancellationToken cancellationToken = default)
{
ArgumentException.ThrowIfNullOrEmpty(key);
ArgumentNullException.ThrowIfNull(initialStateFactory);
// Cannot rely on `value is not null` because T? on an unconstrained generic
// parameter does not become Nullable<T> for value types — the null check is
// always true for types like int. Instead, check key existence directly.
if (this.HasStateKey(key, scopeName))
{
T? value = await this.ReadStateAsync<T>(key, scopeName, cancellationToken).ConfigureAwait(false);
if (value is not null)
{
return value;
}
}
T initialValue = initialStateFactory();
await this.QueueStateUpdateAsync(key, initialValue, scopeName, cancellationToken).ConfigureAwait(false);
return initialValue;
}
/// <inheritdoc/>
public ValueTask<HashSet<string>> ReadStateKeysAsync(
string? scopeName = null,
CancellationToken cancellationToken = default)
{
string scopePrefix = GetScopePrefix(scopeName);
int scopePrefixLength = scopePrefix.Length;
HashSet<string> keys = new(StringComparer.Ordinal);
bool scopeCleared = scopeName is null
? this.ClearedScopes.Contains(DefaultScopeName)
: this.ClearedScopes.Contains(scopeName);
// Start with keys from initial state (skip if scope was cleared)
if (!scopeCleared)
{
foreach (string stateKey in this._initialState.Keys)
{
if (stateKey.StartsWith(scopePrefix, StringComparison.Ordinal))
{
keys.Add(stateKey[scopePrefixLength..]);
}
}
}
// Merge local updates: add if non-null, remove if null (deleted)
foreach (KeyValuePair<string, string?> update in this.StateUpdates)
{
if (!update.Key.StartsWith(scopePrefix, StringComparison.Ordinal))
{
continue;
}
string key = update.Key[scopePrefixLength..];
if (update.Value is not null)
{
keys.Add(key);
}
else
{
keys.Remove(key);
}
}
return ValueTask.FromResult(keys);
}
/// <inheritdoc/>
public ValueTask QueueStateUpdateAsync<T>(
string key,
T? value,
string? scopeName = null,
CancellationToken cancellationToken = default)
{
ArgumentException.ThrowIfNullOrEmpty(key);
string scopeKey = GetScopeKey(scopeName, key);
this.StateUpdates[scopeKey] = value is null ? null : SerializeState(value);
return default;
}
/// <inheritdoc/>
public ValueTask QueueClearScopeAsync(
string? scopeName = null,
CancellationToken cancellationToken = default)
{
this.ClearedScopes.Add(scopeName ?? DefaultScopeName);
// Remove any pending updates in this scope (snapshot keys to allow removal during iteration)
string scopePrefix = GetScopePrefix(scopeName);
foreach (string key in this.StateUpdates.Keys.ToList())
{
if (key.StartsWith(scopePrefix, StringComparison.Ordinal))
{
this.StateUpdates.Remove(key);
}
}
return default;
}
/// <inheritdoc/>
public IReadOnlyDictionary<string, string>? TraceContext => null;
/// <inheritdoc/>
public bool ConcurrentRunsEnabled => false;
private static string GetScopeKey(string? scopeName, string key)
=> $"{GetScopePrefix(scopeName)}{key}";
/// <summary>
/// Checks whether the given key exists in local updates or initial state,
/// respecting cleared scopes.
/// </summary>
private bool HasStateKey(string key, string? scopeName)
{
string scopeKey = GetScopeKey(scopeName, key);
if (this.StateUpdates.TryGetValue(scopeKey, out string? updated))
{
return updated is not null;
}
string normalizedScope = scopeName ?? DefaultScopeName;
if (this.ClearedScopes.Contains(normalizedScope))
{
return false;
}
return this._initialState.ContainsKey(scopeKey);
}
/// <summary>
/// Returns the key prefix for the given scope. Scopes partition shared state
/// into logical namespaces, allowing different workflow executors to manage
/// their state keys independently. When no scope is specified, the
/// <see cref="DefaultScopeName"/> is used.
/// </summary>
private static string GetScopePrefix(string? scopeName)
=> scopeName is null ? $"{DefaultScopeName}:" : $"{scopeName}:";
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Serializing workflow state types.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Serializing workflow state types.")]
private static string SerializeState<T>(T value)
=> JsonSerializer.Serialize(value, DurableSerialization.Options);
[UnconditionalSuppressMessage("AOT", "IL3050", Justification = "Deserializing workflow state types.")]
[UnconditionalSuppressMessage("Trimming", "IL2026", Justification = "Deserializing workflow state types.")]
private static ValueTask<T?> DeserializeStateAsync<T>(string? json)
{
if (json is null)
{
return ValueTask.FromResult<T?>(default);
}
return ValueTask.FromResult(JsonSerializer.Deserialize<T>(json, DurableSerialization.Options));
}
}
@@ -0,0 +1,35 @@
// Copyright (c) Microsoft. All rights reserved.
using System.Diagnostics;
using Microsoft.Agents.AI.Workflows;
using Microsoft.DurableTask;
namespace Microsoft.Agents.AI.DurableTask.Workflows;
/// <summary>
/// Event raised when a durable workflow fails.
/// </summary>
[DebuggerDisplay("Failed: {ErrorMessage}")]
public sealed class DurableWorkflowFailedEvent : WorkflowEvent
{
/// <summary>
/// Initializes a new instance of the <see cref="DurableWorkflowFailedEvent"/> class.
/// </summary>
/// <param name="errorMessage">The error message describing the failure.</param>
/// <param name="failureDetails">The full failure details from the Durable Task runtime, if available.</param>
public DurableWorkflowFailedEvent(string errorMessage, TaskFailureDetails? failureDetails = null) : base(errorMessage)
{
this.ErrorMessage = errorMessage;
this.FailureDetails = failureDetails;
}
/// <summary>
/// Gets the error message describing the failure.
/// </summary>
public string ErrorMessage { get; }
/// <summary>
/// Gets the full failure details from the Durable Task runtime, including error type, stack trace, and inner failure.
/// </summary>
public TaskFailureDetails? FailureDetails { get; }
}

Some files were not shown because too many files have changed in this diff Show More