an AI-driven network emulation agent that mimics real network device behavior using LangChain, LangGraph, and GPT-based LLMs. The goal is to provide a transparent experience where CLI and API interactions feel like a real network device.
Core Components
Responder Agent
Processes user queries such as:
CLI Command: show interfaces
API Request: GET /restconf/data/ietf-interfaces:interfaces
State Agent
️ Tracks the emulator’s state to ensure responses reflect real-time configuration changes.
Execution Interfaces (Management Plane)
SSH Server – Built with Paramiko, allowing users to execute Ansible playbooks seamlessly.
REST API Server – Developed using FastAPI, supporting Python or Terraform execution via RESTCONF.
AI Agent Integration – Wraps around the management plane to handle user commands intelligently, making automation interactions appear transparent and realistic, just like a real device.
AI Stack Used
LangChain + LangGraph – For structured orchestration of command execution.
GPT-based LLMs – For generating realistic CLI and RESTCONF API responses.
Challenges Faced
State Accuracy – Ensuring responses align with real-time config changes.
Realistic Response Timing – Simulating device processing delays accurately.
Precision in CLI & API Responses – Making outputs match real network devices.
RAG with cisco if it is not illegal docs & Prompt Engineering – Fixing retrieval-augmented generation inconsistencies.
