Agentic RAG → NETCONF XML → Orchestrator‑Worker DAG execution with a NiceGUI frontend, powered by Microsoft Agent Framework (MAF) 1.2.0.

This project demonstrates an Agentic RAG–driven network automation workflow using the Microsoft Agent Framework (MAF) 1.2.0.
Natural‑language intent is converted into NETCONF XML, executed through a two‑task DAG, and visualized through a NiceGUI web interface.

It is designed as a minimal, reproducible demo showing how AI‑assisted intent, NETCONF automation, and orchestrator‑worker patterns can be combined for modern network operations.

Demonstration: Multi-Task DAG Execution via MAF 1.2.0 Watch how a simple intent ("Delete VLAN 50 and add VLAN 100") is autonomously handled by MAF agents, with full visibility through NiceGUI and JupyterHub.

• NiceGUI UI
• Natural‑language input
• RAG‑generated NETCONF XML
• DAG execution (2 tasks)
• Before/after CLI output via JupyterHub

• Converts natural‑language intent into structured NETCONF XML
• Uses a local FAISS vector store (excluded from repo)

Demo workflow:

• Task 1: Delete VLAN 50
• Task 2: Add VLAN 100

• Two‑task DAG executed by MAF
• Clear separation of orchestration and execution
• Ideal for multi‑step network workflows

• Input natural‑language intent
• Preview generated XML
• Trigger DAG execution
• Display results

• Shows before/after device state
• Makes the automation process transparent

Natural Language Input
│
▼
Agentic RAG
│
▼
NETCONF XML Generation
│
▼
MAF Orchestrator
│
▼
DAG (2 Tasks)
├── Task 1: Delete VLAN 50
└── Task 2: Add VLAN 100
│
▼
Network Device

git clone https://github.com/hidemi-k/maf-netconf-rag-gui.git
cd maf-netconf-rag-gui

pip install -r requirements.txt

cp config.ini.example config.ini

python maf_netconf_rag_gui.py

Navigate to http://localhost:8080 in your browser.
You will be automatically redirected to http://localhost:8080/netconf_rag.

Note: On startup, the following warning from the BERT model loader can be safely ignored:

embeddings.position_ids | UNEXPECTED

This is expected behavior when loading BAAI/bge-large-en-v1.5 and does not affect functionality.

maf-netconf-rag-gui/
 ├── maf_netconf_rag_gui.py      # Main NiceGUI + RAG + NETCONF app
 ├── config.ini.example          # Example configuration
 ├── requirements.txt            # Python dependencies required for the UI, RAG engine, and NETCONF automation
 ├── LICENSE                     # MIT License
 ├── .gitignore                  # Includes faiss_db/ exclusion
 ├── faiss_db/ (ignored)         # Local vector store
 └── .ipynb_checkpoints/         # Jupyter artifacts

• Python 3.10+
• NiceGUI
• FAISS
• Microsoft Agent Framework (MAF) 1.2.0
• NETCONF-capable network device
• JupyterHub (optional)

This project is licensed under the MIT License. See the LICENSE file for details.

• Microsoft Agent Framework (MAF) team
• NiceGUI project
• Network automation community