Skip to content
Agents tracked: 258 Downloads (7d): 219M up 6.1% GitHub stars: 5.5M VS Code installs: 148M Releases (7d): 293 Agent status: 2 with issues Updated Oct 7, 2026
AgentGid

Semantic Kernel changelog: what's new each month

Every stable Semantic Kernel release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.

October 2026 · so far

2 releases: 1.45.0 → 1.81.0

October brought updates to both .NET and Python components, with dependency upgrades, breaking changes to authentication and HTTP defaults, and improvements to file handling and vector filtering.

Action needed
  • OpenAPI HTTP client defaults changed in .NET
  • Unsupported Copilot Studio agent service auth mode removed from Python

Highlights

  • Breaking change to OpenAPI HTTP client defaults in .NET requires configuration review
  • Removed unsupported Copilot Studio agent service auth mode in Python
  • Improved Milvus vector database filtering performance
  • Enhanced file handling in the FileIOPlugin
  • Updated Gemini connector to respect function choice behavior lists

Improved

  • Milvus filtering logic in vector database operations
  • File handling in FileIOPlugin for .NET
  • Plugin path validation alignment
  • FunctionChoiceBehavior function list support in Gemini connector
  • SessionsPythonPlugin reliability

Fixed

  • SessionsPythonPlugin bug
  • Plugin path validation consistency

September 2026

1 release: 1.80.1

September brought maintenance updates and dependency upgrades to keep Semantic Kernel current with the latest platform versions.

Highlights

  • Updated .NET SDK to version 10.0.303
  • Bumped OpenAPI dependencies to latest versions
  • Removed retired OpenAI Assistants integration tests

Improved

  • Refactored redirection tests to avoid race conditions
  • Updated AWS credential task for GitHub Actions Node.js 20 retirement

Fixed

  • Fixed typos in codebase

August 2026

3 releases: 1.79.0 → 1.80.0

August brought stability improvements and bug fixes across both .NET and Python implementations of Semantic Kernel. The month focused on refining existing functionality and addressing edge cases in vector storage and tool handling.

Action needed
  • Python: MCP tool approval callback added for Azure AI Agent (may require implementation in existing code).

Highlights

  • Fixed CosmosDB vector store bug affecting data operations.
  • OpenAPI server variable values are now properly encoded.
  • Gemini connector now correctly honors function choice behavior settings.
  • MCP tool approval callback added for Azure AI Agent in Python.
  • MCP tools with colliding normalized names are now properly skipped to prevent conflicts.

New

  • MCP tool approval callback for Azure AI Agent in Python.
  • x5t certificate thumbprint hashing documentation for Copilot Studio agent.

Improved

  • OpenAPI HTTP client defaults updated.
  • Gemini connector now honors FunctionChoiceBehavior function list.
  • Vector store samples migrated to CommunityToolkit.VectorData packages.
  • CodeQL false positives suppressed in internal HTTP utility.
  • Shell handling hardened in dotnet-format workflow.

Fixed

  • CosmosDB vector store bug.
  • OpenAPI server variable value encoding.
  • MCP tool name collision handling in Python.

Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. Semantic Kernel: pricing, features and alternatives · All changelogs