This folder centralizes documentation for the Microsoft Agent Framework (MAF) application suite in this repository. Each reference solution highlights a different orchestration pattern, user experience, and integration approach that you can reuse when crafting your own MAF-powered products.This documentation describes the portfolio of reference applications that we have built on top of the Microsoft Agent Framework (MAF). Each solution demonstrates a different orchestration pattern, user experience, and deployment approach that can be reused when assembling your own MAF-powered product.
-
Multi-stage workflows combining sequential, concurrent, and YAML-defined execution modes- Multi-stage research workflows that mix sequential, concurrent, and hand-off execution modes
-
Persistent research history and artifact storage backed by Cosmos DB and Azure Storage- Research persistence, artifact storage, and YAML-driven workflow definitions
-
Full-stack reference for planners, agents, and tool adapters running on MAF- End-to-end example of integrating MAF planners, agents, and tool adapters
-
Quick links: README, Backend, Frontend, Quickstart- Quick links: README, Backend, Frontend, Quickstart
-
Audio, video, and PDF processing with Azure Speech and Document Intelligence- Multimodal analysis that blends image, document, and structured data interpretation
-
Step approvals, execution monitor, and Cosmos-backed history for replayable sessions- Task history with replayable runs and OpenTelemetry-based diagnostics
-
Demonstrates MCP-enabled enrichment and Azure OpenAI powered analytics- Demonstrates MCP-enabled enrichment for Vision and Retrieval workflows
-
Quick links: README, Backend, Frontend, Architecture- Quick links: README, Backend, Frontend, Architecture
-
Real-time meeting intelligence: transcription, summarization, action items, and sentiment insights- Real-time meeting intelligence with transcription, sentiment scoring, recommendations, and summaries
-
Cosmos DB history, Application Insights telemetry, and Azure AI content safety integration- Cosmos DB history, Application Insights telemetry, and human-in-the-loop validation checkpoints
-
Quick links: README, Backend, Frontend- Azure Speech, OpenAI, Language, and PII protection integrated through MAF orchestration
-
Equity research copilot featuring sequential, concurrent, and group-chat patterns### FinAgent App
-
Financial data adapters for SEC, fundamentals, technicals, and earnings plus Azure OpenAI synthesis- Equity research copilot that applies sequential, concurrent, and group-chat coordination patterns
-
Quick links: README, Backend, Frontend, Docs- Financial data adapters for SEC, earnings calls, fundamentals, and technical analysis
-
Integrated PDF report generation and MCP tool orchestration
-
ReAct-style planner with human approval gating on every agent step
-
Cosmos DB session history, telemetry, and MCP-powered Yahoo Finance integration### FinAgent Dynamic App
-
Quick links: README, Backend, Frontend, Quickstart- Dynamic planning with human approval gates for every agent step
-
Cosmos DB session history, detailed task telemetry, and MCP-powered data acquisition
- Illustrates ReAct-style planners running on top of MAF orchestration primitives
- Quick links: README, Backend, Frontend, Quickstart
-
Sequential, Concurrent, Group Chat, Handoff, and Magentic/Deep Research builders powered by official MAF workflows
-
FastAPI + React experience with execution monitors, run history, and Azure Easy Auth hooks
-
Six orchestrations (Sequential, Concurrent, Group Chat, Handoff, Magentic, Deep Research) running on official MAF builders
- FastAPI backend with real-time agent activity streaming, optional Cosmos DB history, and Azure Easy Auth hooks
- React dashboard featuring execution monitors, run history, and configurable API settings
-
Hackathon Curriculum – Guided learning path, pattern labs, and project backlogs aligned with the current portfolio
-
Deep Research Playbooks – End-to-end walkthroughs for research-centric assistants, including quickstart, patterns, and decision guides
-
Reference App Overview – System blueprint for the Deep Research baseline and how it leverages MAF utilities## Architecture and Pattern Guides
-
Patterns & Builders – Explore practical orchestration examples directly in patterns/README.md
-
MAF Architecture Overview – How the reference apps compose MAF services and shared infrastructure
Getting Started- Pattern Reference – Coordination strategies used across the portfolio with implementation notes
- Microsoft Agent Framework Integration – Integration touchpoints, extension hooks, and best practices when wiring agents
# Clone the repository- **[Hackathon Curriculum](./hackathon/README.md)** – Hands-on labs that teach pattern selection, agent design, and deployment workflows
git clone https://github.com/akshata29/agents.git- **[Reference App Overview](./reference-app/overview.md)** – Component breakdown for the Deep Research baseline
cd agents
## Getting Started
# Pick an application to explore
cd deep_research_app/backend```bash
# Clone the repository
# Create an isolated environment and install dependenciesgit clone https://github.com/akshata29/agents.git
python -m venv .venvcd agents
.venv\Scripts\Activate.ps1
pip install -r requirements.txt# Pick an application to explore
cd deep_research_app/backend
# Launch the backend (follow each app README for complete setup)
uvicorn app.main:app --reload --port 8000# Create an isolated environment and install dependencies
```python -m venv .venv
.venv\\Scripts\\Activate.ps1
- Frontend projects boot with `npm install && npm run dev` from their respective directoriespip install -r requirements.txt
- Each app folder includes deployment scripts (`deploy.ps1`, Dockerfiles) for Azure App Service or Container Apps
# Launch the backend (follow the app README for full setup)
## Additional Resourcespython app/main.py
-
Deep Research Quickstart- Frontend projects boot with
npm install && npm run devfrom their respective directories -
Deep Research Patterns- Deployment scripts (
deploy.ps1,deploy.bat, Dockerfiles) are available under each app folder
We welcome contributions! See the Contributing Guide for coding standards, branching strategy, and release workflow.- Deep Research Quickstart
We welcome contributions! Please see the Contributing Guide for details on coding standards, branching, and release processes.