Scope durable task IDs by hosted session GUID

[Nathandrake229]

Summary

• read and validate FOUNDRY_AGENT_SESSION_GUID in hosted AgentServer configuration
• scope physical durable multi-turn task IDs by the session incarnation while preserving public conversation_chain_id
• reuse pending, in-progress, or suspended pre-rollout tasks through legacy-ID lookup
• keep one-shot task IDs and TaskManager public session scope unchanged

Motivation

Deleting a hosted session tombstones its durable task documents. Recreating the
same public session name currently derives the same task ID, so task creation can
return HTTP 409 while the prior tombstone remains. Agents Service now supplies a
new system-generated session GUID for each session incarnation.

Compatibility

• Missing environment value preserves the existing public-session derivation.
• Existing active multi-turn chains remain on their legacy task ID.
• Completed or missing legacy tasks create under the GUID-scoped ID.
• Crash recovery continues from the persisted physical task ID.

Validation

• AgentServer Core focused tests: 44 passed
• AgentServer Responses focused tests: 58 passed
• Changed source files: Pylint 10.00/10

Service dependency

Agents Service implementation: Vienna PR 2267233

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
9 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Pull request overview

Scopes hosted durable multi-turn task IDs by session incarnation GUID while preserving public conversation identities and legacy task recovery.

Changes:

• Adds hosted session GUID configuration and propagation.
• Introduces GUID-scoped task IDs with legacy active-task lookup.
• Adds migration tests and documentation.

Reviewed changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
responses/tests/unit/test_task_id.py	Tests GUID-based task scoping.
responses/tests/unit/test_resilient_input.py	Tests GUID serialization.
responses/tests/unit/test_conversation_lock.py	Tests legacy task migration.
responses/tests/unit/test_conversation_chain_id.py	Clarifies public chain identity.
responses/docs/responses-resilience-spec.md	Documents task ID migration.
responses/CHANGELOG.md	Records the fix.
responses/hosting/_task_id.py	Adds private task scope.
responses/hosting/_resilient_orchestrator.py	Selects compatible physical task IDs.
responses/hosting/_resilient_input.py	Persists session GUIDs.
responses/hosting/_execution_context.py	Carries session GUID context.
responses/hosting/_endpoint_handler.py	Reads hosted GUID configuration.
responses/hosting/_chain_id.py	Updates identity documentation.
core/tests/test_config.py	Tests environment parsing and validation.
core/CHANGELOG.md	Records the configuration feature.
core/tasks/_decorator.py	Exposes internal task lookup.
core/_constants.py	Adds the environment variable constant.
core/_config.py	Adds validated session GUID configuration.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 20 out of 20 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 20 out of 20 changed files in this pull request and generated 1 comment.

Suppressed comments (3)

sdk/agentserver/azure-ai-agentserver-core/CHANGELOG.md:3

• This new release entry is not reflected in the package version: pyproject.toml:61 reads the dynamic version from core/_version.py, which still declares 2.1.0. Consequently a core wheel built from this PR is still version 2.1.0 and cannot satisfy the Responses package's new azure-ai-agentserver-core>=2.2.0b1 requirement. Update _version.py to 2.2.0b1 alongside this changelog.

## 2.2.0b1 (Unreleased)

sdk/agentserver/azure-ai-agentserver-responses/CHANGELOG.md:3

• The Responses package still builds as version 2.1.0: pyproject.toml:35-36 sources its dynamic version from responses/_version.py, which was not updated. This makes the new 2.2.0b1 changelog section disagree with the wheel metadata and would attempt to publish the changes under the already-released version. Update _version.py to 2.2.0b1.

## 2.2.0b1 (Unreleased)

sdk/agentserver/azure-ai-agentserver-responses/azure/ai/agentserver/responses/hosting/_task_id.py:54

• The new private scope makes the existing return documentation inaccurate: when task_session_id differs from session_id, the result is explicitly not the handler-facing conversation chain ID (as the new tests assert). Update the return clause to state that equality holds only when the private scope is omitted or equals the public session scope.

    :keyword task_session_id: Private task namespace. Defaults to
        ``session_id`` for compatibility.
    :paramtype task_session_id: str | None

[Copilot]

Pull request overview

Copilot reviewed 22 out of 22 changed files in this pull request and generated no new comments.

[github-actions]

[Pilot] PR Pipeline Failure Analysis

What failed

Pipeline: python - pullrequest (Build Analyze) — Build ID 6752307
Stage/Job: Build Analyze → verifysdist
Packages: azure-ai-agentserver-core and azure-ai-agentserver-responses (both version 2.2.0b1)

Both packages failed verifysdist with the same root cause: the Development Status classifier is set to 5 - Production/Stable, but the package version (2.2.0b1) is a beta release and must use 4 - Beta. Per Azure SDK policy, a beta version must have is_stable = false in pyproject.toml and the classifier Development Status :: 4 - Beta.

There is also a secondary issue for azure-ai-agentserver-responses: a pip-resolution error (Could not find a version that satisfies the requirement azure-ai-agentserver-core<2.3.0,>=2.2.0b1) when resolving from the public feed, because azure-ai-agentserver-core==2.2.0b1 has not yet been published. This resolved via local source install, but indicates the version 2.2.0b1 is not yet on the feed.

Relevant pipeline output

[ERROR] azure-sdk-tools: azure-ai-agentserver-core has version 2.2.0b1 and is a beta release,
but has development status 'Development Status :: 5 - Production/Stable'.
Expected 'Development Status :: 4 - Beta' ONLY.
[ERROR] azure-sdk-tools: azure-ai-agentserver-core failed sdist verification. Check outputs above.

[ERROR] azure-sdk-tools: azure-ai-agentserver-responses has version 2.2.0b1 and is a beta release,
but has development status 'Development Status :: 5 - Production/Stable'.
Expected 'Development Status :: 4 - Beta' ONLY.
[ERROR] azure-sdk-tools: azure-ai-agentserver-responses failed sdist verification. Check outputs above.

PACKAGE                                                            CHECK        STATUS  DURATION(s)
---------------------------------------------------------------------------------------------------
sdk/agentserver/azure-ai-agentserver-core                         verifysdist  FAIL(1)        14.66
sdk/agentserver/azure-ai-agentserver-responses                    verifysdist  FAIL(1)        14.69

Total checks: 2 | Failed: 2 | Worst exit code: 1

Recommended next steps

• In sdk/agentserver/azure-ai-agentserver-core/pyproject.toml: change "Development Status :: 5 - Production/Stable" to "Development Status :: 4 - Beta" in the classifiers list, and set is_stable = false.
• In sdk/agentserver/azure-ai-agentserver-responses/pyproject.toml: apply the same classifier and is_stable fix.
• See https://aka.ms/ci-fix

Automated fix: Requested

Generated by Pipeline Analysis Next Steps · 47.7 AIC · ⌖ 6.56 AIC · ⊞ 6K · ◷

[Copilot]

Pull request overview

Copilot reviewed 23 out of 23 changed files in this pull request and generated no new comments.

[Copilot]

Pull request overview

Copilot reviewed 23 out of 23 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 22 out of 22 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 23 out of 23 changed files in this pull request and generated no new comments.