agent-framework-core
Microsoft Agent Framework for building AI Agents with Python. This is the core package that has all the core abstractions and implementations.
What it is and what it does
Agent Framework Core is Microsoft's Python library for building and orchestrating AI agents powered by large language models. It provides abstractions for creating individual agents with custom instructions and tools, then composing them into multi-agent systems using patterns like sequential workflows, concurrent execution, group chat, and handoff protocols. The framework handles LLM communication, function calling, message routing, and telemetry.
You use it by instantiating agents with configuration, optionally attaching custom Python functions as tools, then either running agents directly or coordinating multiple agents through orchestration patterns. It supports both in-process and distributed execution, multimodal inputs (text and vision), and integrates with opentelemetry-api for observability. Core dependencies include msgspec, pydantic, typing-extensions, python-dotenv, and opentelemetry-api.
Use it for:
- Build a single agent with custom tools to answer questions and call functions on behalf of users.
- Orchestrate multiple agents in sequence to iteratively refine outputs based on feedback.
- Run multiple specialized agents concurrently to parallelize work on independent subtasks.
- Implement group chat where multiple agents collaborate and discuss to reach consensus.
- Deploy agents in distributed environments for scalable, fault-tolerant multi-agent applications.
Worth the install?
AI-flagged interpretation of the facts on this page — verify before relying
Build, orchestrate, and deploy AI agents and multi-agent systems with support for multiple LLM providers, function calling, and orchestration patterns.
Yes. Active, well-maintained framework with strong community adoption (12803 stars, top 5000 PyPI packages). MIT license, low install friction, no known vulnerabilities, and comprehensive multi-agent orchestration capabilities. Install if you need to build or coordinate AI agents; skip if you only need direct LLM chat without agent abstractions.
Install
agent-framework-core on PyPI
pip
pip install agent-framework-coreuv
uv add agent-framework-corepoetry
poetry add agent-framework-coreInstalling agent-framework-core
Before you install
Low friction installation with a pure-Python wheel. Active maintenance with recent release and strong community signal (12803 stars). Supports Python 3.10 through 3.14.
License in practice
MIT license (permissive) allows unrestricted use, modification, and distribution in commercial and private projects with minimal obligations.
Quickstart
pip install agent-framework-core
import asyncio
from agent_framework_core import Agent
agent = Agent(
instructions="You are a helpful assistant."
)
result = asyncio.run(agent.run("Hello"))
print(result)
Requires Python 3.10+. Requires LLM provider credentials (e.g., OPENAI_API_KEY) set via environment variables or configuration.
Verify before relying
- Which LLM providers (OpenAI, Foundry, Anthropic) are included in core versus optional integration packages
- Performance characteristics and latency overhead of the orchestration layer
- Whether distributed execution requires additional infrastructure or dependencies beyond the core package
Package facts
| License | not declared (permissive) |
| Python support | supports the current Python release (>=3.10) |
| Install friction | low — pure-Python wheel |
| Runtime dependencies | 5 — msgspec, typing-extensions, pydantic, python-dotenv, opentelemetry-api |
| Maintenance | actively maintained — 0 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 2,281,417/month — #3,165 on PyPI (30-day window, as of 2026-08-14) |
| Known vulnerabilities | none known (OSV.dev, checked 2026-08-14) |
Evidence: agent_framework_core-1.14.0-py3-none-any.whl
Tags
More Application Frameworks packages
FastAPI is a Python web framework for building…
permissive · top 100 on PyPI
annotated-docProvides a way to document function parameters,…
permissive · top 100 on PyPI
textualTextual is a Python framework for building…
permissive · top 100 on PyPI
typerTyper builds command-line applications from…
permissive · top 1,000 on PyPI
mcpBuild and connect to Model Context Protocol…
permissive · top 1,000 on PyPI
WerkzeugWerkzeug is a WSGI utility library providing…
permissive · top 1,000 on PyPI
agent-frameworkBuilds and orchestrates AI agents powered by…
permissive · top 5,000 on PyPI
agent-framework-orchestrationsProvides high-level builder patterns for…
permissive · top 15,000 on PyPI
ai-parrotAI-Parrot is an async-first Python framework…
permissive · top 15,000 on PyPI
marvinMarvin is a Python framework for building AI…
permissive · top 15,000 on PyPI
semantic-kernelSemantic Kernel is a Python SDK for building AI…
permissive · top 5,000 on PyPI
agent-framework-foundry-localIntegrates Microsoft Agent Framework with…
permissive · top 15,000 on PyPI
agent-framework-claudeIntegrates Claude with Microsoft Agent…
permissive · top 15,000 on PyPI
agent-framework-ollamaIntegrates Ollama LLM provider support into…
permissive · top 15,000 on PyPI
agent-framework-anthropicConnects Agent Framework applications to…
permissive · top 15,000 on PyPI
agent-framework-devuiProvides a lightweight web UI and…
permissive · top 15,000 on PyPI