Implementation:Microsoft Semantic kernel Python Pyproject Configuration

Knowledge Sources	Microsoft_Semantic_kernel
Domains	Python, Build_Configuration, Package_Management
Last Updated	2026-02-11 00:00 GMT

Overview

Python project configuration and build metadata file (pyproject.toml) for the semantic-kernel Python SDK package, defining dependencies, optional extras, dev tools, and build system settings.

Description

This pyproject.toml file is the central build and project configuration for the Semantic Kernel Python SDK. It defines the package metadata, core dependencies, optional dependency groups for various AI providers and vector stores, development tooling configuration, and the build system.

Project metadata:

• Package name: semantic-kernel
• Python requirement: >=3.10
• License: MIT
• Development status: Production/Stable
• Build system: Flit (flit-core >= 3.9,<4.0)
• Version: Dynamically read from __init__.py

Core dependencies include:

• azure-ai-projects, azure-ai-agents, azure-identity - Azure service integrations
• pydantic (>=2.0), pydantic-settings - Data validation and settings
• openai (>=1.98.0,<2) - OpenAI API client
• opentelemetry-api, opentelemetry-sdk - Observability and tracing
• numpy - Embedding computations (version varies by Python version)
• jinja2, pybars4 - Template rendering
• openapi_core, prance - OpenAPI/Swagger support
• mcp - Model Context Protocol support

Optional dependency groups (extras):

• AI Providers: anthropic, aws (boto3), google (google-genai, vertex AI), hugging_face (transformers, sentence-transformers), mistralai, ollama, onnx
• Vector Stores: azure (azure-search, cosmos), chroma, milvus, mongo, pinecone, postgres (psycopg), qdrant, redis, usearch, weaviate, faiss, oracledb
• Other: autogen, copilotstudio, dapr, notebooks, pandas, realtime, sql

Development tools:

• pytest (~=8.2) with asyncio, coverage, xdist, and timeout plugins
• mypy (>=1.10) for static type checking
• ruff (~=0.9) for linting and formatting with Google pydocstyle convention
• pre-commit (~=3.7) for git hooks

Usage

This file is used by Python package managers (pip, uv, flit) to install the semantic-kernel package and its dependencies. Developers install the base package with pip install semantic-kernel or specific extras with pip install semantic-kernel[google,redis]. Contributors use the dev dependency group for testing and linting. The file is also the source of truth for tool configurations (ruff, pytest, bandit).

Code Reference

Source Location

• Repository: Microsoft_Semantic_kernel
• File: python/pyproject.toml
• Lines: 1-262

Signature

[project]
name = "semantic-kernel"
description = "Semantic Kernel Python SDK"
authors = [{ name = "Microsoft", email = "SK-Support@microsoft.com"}]
readme = "README.md"
dynamic = ["version"]
requires-python = ">=3.10"
license = {file = "LICENSE"}
urls.homepage = "https://learn.microsoft.com/en-us/semantic-kernel/overview/"
urls.source = "https://github.com/microsoft/semantic-kernel/tree/main/python"

dependencies = [
    "azure-ai-projects ~= 1.0.0b12",
    "pydantic >=2.0,!=2.10.0,!=2.10.1,!=2.10.2,!=2.10.3,<2.12",
    "openai >= 1.98.0,<2",
    "opentelemetry-api ~= 1.24",
    "numpy >= 1.25.0; python_version < '3.12'",
    "jinja2 ~= 3.1",
    "mcp>=1.26.0",
]

Import

# Install the base semantic-kernel package
pip install semantic-kernel

# Install with specific optional dependencies
pip install semantic-kernel[google,redis,anthropic]

# Install with all development dependencies
pip install semantic-kernel
pip install --group dev

I/O Contract

Inputs

Name	Type	Required	Description
Python version	runtime	yes	Python >=3.10 required to install and use the package
extras	string list	no	Optional dependency groups to install (e.g., "google", "redis", "anthropic")

Outputs

Name	Type	Description
semantic-kernel package	Python package	Installed semantic_kernel module with all core dependencies
optional integrations	Python packages	Additional provider-specific or store-specific packages based on selected extras
dev tools	Python packages	pytest, mypy, ruff, and other development tools when dev group is installed

Usage Examples

Installing with Specific Extras

# Install for use with Google AI and Qdrant vector store
pip install semantic-kernel[google,qdrant]

# Install for Azure ecosystem (AI Inference, Cosmos, Search)
pip install semantic-kernel[azure]

# Install for Hugging Face local models
pip install semantic-kernel[hugging_face]

Running the Test Suite

# Install development dependencies
cd python
pip install -e .
pip install --group dev

# Run tests with pytest (configured in pyproject.toml)
pytest tests/ -ra -q

Running Ruff Linter

# Lint the project (configuration from pyproject.toml)
ruff check python/semantic_kernel/

# Format code
ruff format python/semantic_kernel/

Related Pages

• Principle:Microsoft_Semantic_kernel_Python_SDK_Package_Configuration