CortexCMMS MCP Server for CMMS: Connect Industrial Systems to Agentic AI

CortexCMMS MCP — Connect Industrial Systems to Agentic AI

Architecture Overview

CortexCMMS MCP is designed as an integration and action layer that connects industrial and enterprise systems with the Hexa AI Agent Ecosystem. The architecture provides a structured pathway for bringing operational context into AI-driven workflows and, where configured and authorized, turning AI decisions into operational actions inside CortexCMMS.

The architecture separates four major layers: connected systems, the CortexCMMS MCP integration layer, Hexa AI agents, and the CortexCMMS operational layer.

1Connected SystemERP, MES, SCADA, PLC, IoT, Historian, CMMS / EAM, APIs

2CortexCMMS MCP Integration LayerConnectors, APIs, data mapping, security & governance

3Hexa AI Agent EcosystemUnderstand, analyze, prioritize, recommend, act

4CortexCMMS Operational LayerWork orders, scheduling, asset health, compliance

Connected Systems

CortexCMMS MCP can be positioned between CortexCMMS/Hexa AI and the systems where operational data and enterprise context reside. Typical integration targets include:

ERP

MES

SCADA

PLC / Automation

IoT & Sensors

Historian

Existing CMMS / EAM

APIs & Databases

The objective is to reduce the need for technicians, planners, engineers, and operations leaders to manually move information between disconnected applications.

CortexCMMS MCP Integration Layer

The MCP layer provides the integration boundary between external systems and CortexCMMS's agentic workflows. In an implementation, this layer can handle connectors, APIs, data mapping, authorization, security controls, and governance.

Connectors

APIs

Data Mapping

Security & Governance

Hexa AI Agent Ecosystem

Hexa AI is the agentic layer within CortexCMMS. Instead of treating maintenance software only as a system for entering and retrieving records, the agent layer is designed to reason over connected context and coordinate multi-step workflows.

UnderstandsAnalyzePrioritizeRecommendCreateAssignFollow UpDocument

Example sequence

An equipment notification can be contextualized with asset history and sensor data, analyzed for maintenance relevance, prioritized, converted into a recommended action, and then routed through the configured work-order and approval workflow.

CortexCMMS Operational Layer

The operational layer is where maintenance and asset-management outcomes are represented and managed. Potential capabilities include:

Equipment Intelligence

Work Orders

Maintenance Scheduling

Audit & Compliance Reports

Asset Health

Inventory & Resources

Agentic Maintenance Workflow

1ConnectRetrieve relevant information from authorized systems.

2ContextualizeCombine asset, production, maintenance, and operational context.

3AnalyzeEvaluate conditions, history, alarms, work history, or other available evidence.

4DecideDetermine the appropriate next step within configured business rules and agent instructions.

5ActInvoke authorized tools or CortexCMMS workflows such as creating or updating work orders.

6VerifyConfirm status, outcome, or required follow-up.

7DocumentCapture the resulting action, evidence, and maintenance record.

Example MCP Resources and Tools

Illustrative resource patternsresources

asset://pump/P-102

asset://motor/M-204

asset://workorder/WO-18452

asset://equipment/P-102/history

asset://site/plant-01

asset://production/line-4

Illustrative tool patternstools

get_asset()

get_asset_history()

get_sensor_data()

get_alarm_history()

create_work_order()

update_work_order()

assign_work_order()

schedule_maintenance()

create_inspection()

get_pm_schedule()

update_pm_schedule()

get_inventory()

check_spare_parts()

generate_compliance_report()

generate_audit_report()

Read-oriented accessTools that create side effects

Security & Governance

Authentication and authorization for connected systems and MCP clients.

Role-based access control (RBAC) and least-privilege tool access.

Auditable records of data access and operational actions.

Human approval or human-in-the-loop controls for workflows that require authorization.

Separation between read-oriented data access and tools that can create side effects.

Operational governance for agent instructions, tool permissions, and integration mappings.

Developer Integration Model

The developer-facing concept is to expose the systems and capabilities required by an AI application through standardized MCP primitives. The current MCP ecosystem defines servers that can expose tools, resources, and prompts to connected AI applications. This makes MCP useful as a protocol boundary between AI applications and the systems where enterprise or operational data and tools live.

For CortexCMMS, the architectural goal is to use this boundary to make industrial context accessible to Hexa AI while keeping system access governed by the enterprise's integration, identity, security, and authorization model.

Hexa AIAI application inside CortexCMMS

CortexCMMS MCP ServerProtocol boundary, governed by your identity and authorization model.

toolsresourcesprompts

Enterprise & operational systemsWhere enterprise or operational data and tools live

Developer Integration Model

See how CortexCMMS MCP connects your systems to Hexa AI.