gpaasch's picture
Update README.md
091333d verified

A newer version of the Gradio SDK is available: 6.1.0

Upgrade
metadata
title: Network Change Simulator
emoji: 🐠
colorFrom: pink
colorTo: red
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
pinned: false
license: fair-noncommercial-research-license
short_description: MCP tool to simulate network changes before deployment
tags:
  - building-mcp-track-enterprise

tag: building-mcp-track-enterprise

video: https://www.youtube.com/watch?v=fRcM7jvANt8

social media post: https://www.linkedin.com/posts/grahampaasch_network-change-simulator-enterprise-mcp-activity-7396053837951766528-3A_E?utm_source=social_share_send&utm_medium=member_desktop_web&rcm=ACoAABO-OuYBtWdvfyVGnYlWMX4qmiNHrtFV06Q

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Network Change Simulator (NCS)

Network Change Simulator (NCS) is an MCP server and Hugging Face Space that lets an infra engineer safely trial network changes before touching production. It simulates VLAN / interface / BGP neighbor changes against our MAESTRO lab (GNS3 + Ansible/Nornir/pyATS), runs targeted pre- and post-change health checks, and returns a 0–100 risk score with an explanation in a few seconds. Lightning mode gives a fast, MCP-driven risk assessment for the included presets; Full mode will handle slower, deeper validation later.

Track & category

This project is submitted to Track 1 – Building MCP, in the Enterprise MCP Servers category (building-mcp-track-enterprise). It is designed as a serious infrastructure MCP server rather than a consumer or creative app.

Why this exists

Who it’s for
Infra/SRE engineers building or expanding AI/data-center campuses—hundreds of racks, ToR switches, and spine–leaf fabrics—who need to know “how risky is this change?” before pushing configs.

What Lightning does
Takes a change preset (VLAN, TOR uplink, BGP neighbor), queries MAESTRO’s health, combines it with change magnitude, and returns a risk score plus pre-/post-check summaries and blast radius text via a single MCP tool and the Space UI.

Why MCP + MAESTRO + HF Space
MCP gives agents/tools a standardized interface, MAESTRO provides a realistic multi-vendor lab, and the Hugging Face Space gives judges/operators a one-click way to see risk without touching production gear.

Available MCP tools

simulate_network_change

  • Inputs
    • change_type: "vlan" | "interface" | "bgp_neighbor"
    • preset_id: preset identifier (e.g. leaf_tor_vlan_stage)
    • mode (optional): "lightning" (implemented) or "full" (placeholder)
  • Outputs
    • risk_score (0–100), risk_level ("low" | "medium" | "high")
    • pre_check_summary[], post_check_summary[]
    • blast_radius_summary, explanation
    • mode, mode_note

How to use

In the Space

  1. Open the Lightning Simulation tab.
  2. Choose a Change type and Scenario preset.
  3. Click Run Lightning risk simulation.
  4. Read the risk score/level, pre/post checks, blast radius, and explanation.

Via MCP clients

  1. Add this snippet to your MCP config:
{
  "network-change-simulator": {
    "sse": {
      "url": "https://hf.space/embed/MCP-1st-Birthday/network-change-simulator/gradio_api/mcp/sse"
    }
  }
}
  1. Call the simulate_network_change tool with the desired preset.
  2. Use the returned risk_score, risk_level, and explanation to decide whether to proceed or escalate.

Under the hood

Network Change Simulator is backed by a live MAESTRO lab: a GNS3 fabric with multi-vendor devices and a library of automation playbooks (Ansible/Nornir/pyATS). Lightning mode simulates those workflows using change presets plus a real MAESTRO health probe (maestro-health-check --quiet) so it can respond in a few seconds. We intentionally keep NCS sponsor-neutral and LLM-optional so it stays reliable, fast, and easy to integrate as an enterprise MCP server; heavier sponsor integrations are reserved for separate agent apps built on top of this server. Full mode is designed to call into those runbooks and deeper validation scripts end-to-end in future iterations.

For deeper details, see docs/RISK_MODEL.md, docs/DEMO_SCRIPT.md, and docs/VIDEO_SCRIPT.md.