Nexus — Enterprise Integration Platform

A self-hosted, sharded integration server that connects enterprise business systems and surfaces what they can't see individually.

---

The Problem

Enterprise systems don't talk to each other. They each hold a piece of the truth — and the gaps between them are where problems hide.

A terminated employee still active in Active Directory. A device enrolled in Intune but untracked in the helpdesk. Fifty laptops inbound from a voluntary retirement program with no system of record for where they go next. A user generating tickets at an unusual rate that nobody notices because the data lives in three different places.

These aren't edge cases. They're the normal state of enterprise IT when systems grow independently and nobody builds the layer that connects them.

Nexus is that layer.

---

What It Does

Nexus polls enterprise systems via their native APIs, normalizes the data, and does two things: surfaces inconsistencies between systems, and aggregates what no single system can report alone.

Each integration is a self-contained shard: independently togglable, independently testable, and designed to be extended without touching anything else. The platform grows with the problem.

Current focus areas:

• Identity — Reconciles user records across Active Directory and Entra ID. Detects status drift, title mismatches, department changes, and name variances before they become compliance findings. Workday HCM integration is in active development.

• ITSM (in development) — Pulls ticket data from BMC Helix to enable operational reporting across users, devices, and issue types Answers questions no single system can: how many tickets has this device generated? What's the incident pattern across a refresh cohort? Which users generate the most escalations?

• Assets (in development) — Correlates device data across Intune, Lansweeper, and BMC Helix to build a complete picture of device assignment history. Built to support PC refresh programs, voluntary retirement collections, and asset lifecycle management.

• Logistics (in development) — Tracks device movement through receiving, staging, deployment, and decommission. Started as a solution for tracking inbound refresh hardware; designed to scale to full lifecycle logistics.

The roadmap: Host in Azure. Expose via Microsoft Copilot agent. Turn multi-system investigations into conversations.

---

Built By

Nathan Castaldi — IT systems and integration practitioner. Built on own time, against real enterprise problems observed in a production environment. Portfolio: linkedin.com/in/nathancastaldi

---

---

Shard Status Board (Traffic Light)

Shard	System(s)	Status	NEXUS Ref	Flag	Standard Gate
identity	Active Directory + Entra ID	🟢 Green	NEXUS-017	ENABLE_IDENTITY	Tool tests passing
workday	Workday HCM	🟡 Yellow	NEXUS-009	ENABLE_WORKDAY	Credentials + live validation pending
audit	Cross-system drift + reporting	🟡 Yellow	NEXUS-018	ENABLE_AUDIT	Verification maturing
itsm	BMC Helix ITSM	🔴 Red	NEXUS-021	ENABLE_ITSM	Stub only
assets	Lansweeper + Intune	🔴 Red	NEXUS-022	ENABLE_ASSETS	Stub only
logistics	FedEx	🔴 Red	NEXUS-023	ENABLE_LOGISTICS	Stub only

Folder Structure

nexus-mcp/
├── src/
│   ├── main.py              # Core Orchestrator — reads flags, loads shards
│   └── shards/              # The 6 Shards (one file = one system domain)
│       ├── __init__.py
│       ├── identity.py      # 🟢 AD & Entra tools
│       ├── workday.py       # 🟡 Worker & Org tools
│       ├── itsm.py          # 🔴 BMC Helix tools
│       ├── assets.py        # 🔴 Lansweeper + Intune tools
│       ├── logistics.py     # 🔴 FedEx tools
│       └── audit.py         # 🟡 Cross-system drift + reporting
├── lib/                     # Low-level system adapters (no tool logic here)
│   ├── config.py
│   ├── ad_adapter.py        # LDAP/AD connection wrapper
│   ├── entra_client.py      # Microsoft Graph (Entra)
│   ├── workday_client.py    # Workday REST + OAuth2
│   ├── helix_client.py      # BMC Helix AR-JWT auth
│   ├── lansweeper_client.py # Lansweeper GraphQL
│   ├── intune_client.py     # Microsoft Graph (Intune)
│   └── fedex_client.py      # FedEx REST + OAuth2
├── tests/
├── .env                     # Feature flags + credentials
├── .env.example             # Template — copy this to .env
└── README.md

---

Architecture: The Shard Contract

Every shard file exposes exactly one function:

def register(mcp: FastMCP) -> None:
    @mcp.tool()
    async def my_tool(...) -> ...:
        """Tool docstring visible to the LLM."""
        ...

The orchestrator (src/main.py) reads feature flags and calls register(mcp) for each enabled shard. No other file is changed to add or remove a shard.

Adding a new shard

1. Create src/shards/my_system.py following the template above.

2. Add the adapter to lib/ if needed.

3. Add one line to src/main.py:

   from shards import my_system
   if _enabled("MY_SYSTEM"):
       my_system.register(mcp)
   

4. Add ENABLE_MY_SYSTEM=true to .env.

Holding pattern

Leave a shard unregistered (or set flag to false) to hold it without breaking anything:

# 🔴 Planned — credentials not yet available
# if _enabled("MY_SYSTEM"):
#     my_system.register(mcp)

---

Local Use and Demo Guide

• See DEMO_GUIDE.md for the full permission matrix, credential configuration, and live-mode setup.
• All credentials live in nexus-mcp/.env — no need to put them in the Claude config.