mcp-rdcopilot-skyspark

Asking a question about a portfolio requires mastering the platform's query language and tag structure; taking over a model built by someone else means guessing what is missing.

This module opens a SkySpark project to a conversational assistant installed on the user's workstation, through MCP (Model Context Protocol), the standard socket through which an assistant connects to a line-of-business application. You write your question in natural language, and the assistant looks for the answer in the portfolio's model — sites, equipment, points, histories and analytics rules. It returns what that project says, with its tags and units, not a textbook answer; it can also create and link records, checking references before writing.

• Query a model without writing its language.
• Understand a model you have taken over: the assistant describes it and flags what is missing.
• Produce the model, not merely query it: records, rules and views.
• Move data between projects and tools, with no script to maintain.
• Bring analytics and the field together: one question crosses servers and stations.

• Model engineering: building a site's model, then its rules and views.
• Taking over a portfolio: describing an unknown project and listing its gaps.
• Model quality: checking tags, references and coverage before delivery.
• Energy monitoring: querying a portfolio's histories and producing a dated status.
• Cross-cutting diagnosis: correlating an operational alarm with field data.

SkySpark 4.x, Python ≥ 3.10

0.20.1

on request

Operate an installation in natural language

See the rangemcp-rdcopilot-niagara