microsoft/content-processing-solution-accelerator v2.1.1

v2.1.1

Repository: microsoft/content-processing-solution-accelerator

Tag: v2.1.1

Published: 2026-06-11T06:06:46Z

Prerelease: no

Release notes:

2.1.1 (2026-06-11)

• chore: dev to main merge (4113db3)
• chore: revert irrelevant docstring/comment changes from upgrade (ce961ad)
• chore: upgrade agent-framework from 1.1.1 to 1.3.0 (5022e34)
• chore(infra): regenerate main.json with bicep 0.43.8 (no semantic change) (1e1fc41)
• Merge branch 'dev' into feature/upgrade-azure-ai-libs (8255859)
• Merge branch 'feature/upgrade-azure-ai-libs' into feature/upgrade-azure-ai-libs-v2 (5eefab5)
• Merge pull request #615 from microsoft/psl-bug46192 (fea3ac8), closes #615 #46192
• Upgrade azure-ai-projects to 2.1.0 and agent-framework to 1.3.0 (746b796)
• fix: align all code with agent-framework 1.1.1 API (793f1b3)
• fix: apply breaking-change fixes for agent-framework 1.1.1 (4e87965)
• fix: migrate all renamed agent-framework symbols for 1.1.1 (b400522)
• fix: pass Message as positional arg to Agent.run() (320a6b3)
• fix: pass start_executor to WorkflowBuilder constructor (803fc11)
• fix: regenerate uv.lock files for agent-framework 1.1.1 (ee4650a)
• fix: remove await from sync ctx.set_state() and fix Role enum usage (9c23281)
• fix: remove unused imports and trailing blank line (lint fixes) (f7c4134)
• fix: replace invalid 'reasoning' dict with 'reasoning_effort' string param (13fe76a)
• fix: replace removed event classes with WorkflowEvent type checks (e73f9f4)
• fix: replace Role.USER with 'user' string in test_input_observer_middleware (96df974)
• fix: skip logprobs for reasoning models in map_handler (83e189a)
• fix: strip temperature/top_p for reasoning models in AgentBuilder (f934b45)
• fix: update agent_framework.azure imports to agent_framework.openai for 1.3.0 (c234316)
• fix: update docstring example to use Agent() constructor pattern (1e66f9e)
• fix: update test patches to use new agent_framework.openai classes (1b14417)
• fix: Upgrade azure-ai-projects and agent-framework (45dd822)
• fix: use executor instances in WorkflowBuilder (1.1.1 API) (1927aa0)
• fix: Workflow.run_stream() -> run(stream=True) in 1.1.1 (369e3f7)
• fix(post-deploy): harden Content Understanding refresh against stale azd env value (23d768c), closes #46192 #46576 #46577
• fix(post-deploy): preserve stderr so transient CLI errors don't masquerade as resource-not-found (0b5d1a7)
• fix(post-deploy): refuse to guess when multiple AIServices accounts exist (f67f5f7)
• refactor: promote _is_reasoning_model and _resolve_model_name to public API (0b7c3f7)
• revert: change agent-framework to 1.1.1, revert all 1.3.0 code changes (77f0171)