Canonical: https://instain.ru/en/ai/microsoft-agent-framework/
Language: en

Microsoft  framework for applications with agents

# Microsoft Agent Framework. 

Workflow graphs with checkpoints for Python and .NET

4.45 / 5    [Directory score](https://instain.ru/en/ai/microsoft-agent-framework/#editorial-rating) [Official website ](https://learn.microsoft.com/agent-framework/)

Process nodes can execute normal code or contact an agent. Connections between nodes determine the route, and control points allow you to save the state of execution.

![Microsoft Guide with review of agents, graphs and placement](https://instain.ru/assets/images/microsoft-agent-framework-screen-5b98afbc0dbb-1200.webp)

Microsoft Guide with review of agents, graphs and placement.  [Open the source ](https://learn.microsoft.com/ru-ru/agent-framework/)

## Where to keep the state

A return request has steps with different costs of failure. You can look up the return policy or retrieve the order number again. Writing a refund request twice creates a duplicate. Microsoft Agent Framework lets you define a workflow graph with separate nodes for the agent, validation, and a regular function. Put checkpoints near steps where relying on process memory would be risky.

## Check on interrupted application

Make two branches for the learning process. In the first, the agent offers a response according to the rules, and the operator allows it to be sent. The second one lacks the order number, so the processing is waiting for clarification. Direct sending to the stub, then stop the process after confirmation and restore it from the control point. Compare the logs before and after the stop. It should show which nodes have already worked and whether the application has reached the imitation of sending.

## What remains in the application code

The framework helps resume the workflow. The application still needs protection against duplicate writes, for example through a stable action key tied to the request number. In a test, delay the tool response and repeat the call from another session. This shows how request state, model messages, and the actual write are kept separate. Before deployment, pin the SDK and model adapter versions because available features depend on the connection.

## Suitable tasks

- Resume processing from a checkpoint

- Stop the agent process before confirmation

Features 

## Strengths

- Graph nodes and state preservation are documented

- Examples for Python and .NET

Things to consider 

## Limitations

- Persisting state does not prevent duplicate operations in external systems

- Features depend on the SDK and model provider you choose

## Selection and conditions

Free access

Yes

Trial period

Not confirmed

No sign-up required

Not confirmed

Russian interface

Not confirmed

Local installation

Yes

API

Not confirmed

“Not confirmed” means there is not enough information. The feature may still be available; check the developer’s terms.

Input formats

Text, Document

Result

Text

### Terms

The open-source framework has no subscription of its own. Model usage and hosting can incur costs. Persisting state does not prevent duplicate operations in external systems.

From task to first result

## Practical example

Task

Resume processing from a checkpoint

What to prepare

Sample customer requests and instructions for routing them between departments.

### Task brief or prompt

For each appeal, offer a category, a brief explanation and the next step according to the instructions. If you don't have enough data, pass it on to the person. Prepare drafts of changes; recording to external systems is allowed only after verification.

### Expected result

Verifiable plan for processing requests and drafts of actions in the configured process.

### How to check the result

- Check unknown categories and empty entry.

- Check connection permissions and ensure the workflow stops before any action requiring approval.

An editorial teaching example of a task brief, not the result of hands-on testing of the service. Check that the required mode is available, that you have rights to the source material and that the intended use of the result is permitted.

Check details

## Official sources

-  [Microsoft Agent Framework documentation ](https://learn.microsoft.com/agent-framework/)

-  [Official Microsoft Repository ](https://github.com/microsoft/agent-framework)

-  [Graph nodes and state persistence ](https://learn.microsoft.com/en-us/agent-framework/concepts/workflows/)

This description is based on the developer’s material. Features, prices and availability may change; check the primary source before use.

About the tool

Developer

Microsoft

Category

[Code](https://instain.ru/en/category/code/)

Access

Use Python or .NET for the graph example. Connect the chosen model and configure workflow deployment separately.

Payment

The open-source framework has no subscription of its own. Model usage and hosting can incur costs.

API

No separate API specified

[Developer terms ](https://learn.microsoft.com/en-us/agent-framework/concepts/workflows/)

Directory score

4.45 / 5 

Features

4.50 / 5 

Workflows

4.20 / 5 

Ecosystem

4.50 / 5 

Documentation

4.60 / 5 

Scores are an editorial guide to the directory. They are not user reviews or the results of personal testing.

[See the methodology ](https://instain.ru/en/methodology/)

Keep exploring

## For similar tasks

[Full category  ](https://instain.ru/en/category/code/)

Code 

###  [Dify](https://instain.ru/en/ai/dify/)

Visual assembly of AI applications with knowledge base, publication and observation

4.53 / 5   Platform  

API availability confirmed

Code 

###  [CrewAI](https://instain.ru/en/ai/crewai/)

Agents with separate roles, tasks, and controlled operating procedures

4.50 / 5   Platform  

API availability confirmed

Code 

###  [LangGraph](https://instain.ru/en/ai/langgraph/)

Agent workflow graphs with checkpoints

4.48 / 5   Status Framework for Agents  

Similar tasks; compare terms, formats and limitations
