EnergyPlus MCP Server

MCP serverOfficial (tool developers) · Pre-1.0

LBNL's MCP server for EnergyPlus: 35 tools to load, inspect, edit and simulate IDF models, run in Docker. By Lawrence Berkeley National Laboratory (LBNL-ETA).

125 starsUpdated 2 days agoGitHub data from 5 Nov 2026

At a glance

Type: MCP server
Works with: EnergyPlus
How it connects: Reads and writes files; the tool isn't needed
Access: Can modify models
Tool versions: EnergyPlus 26.1.0 by default; other releases by rebuilding the Docker image
Operating systems: Windows, macOS, Linux
Needs a EnergyPlus licence: No
Transport: Local (stdio), Remote (HTTP)
AI clients documented: Claude Desktop, VS Code, Cursor
Licence: BSD-3-Clause-LBNL
Language: Python
Release status: pre 1 0

What it can do

Loads and validates IDF files, then lists zones, surfaces, materials, schedules, people, lighting and equipment loads.
Edits occupancy, lighting, equipment and infiltration, adds window films, coatings, output variables and meters, and changes run periods.
Runs EnergyPlus with a weather file and builds interactive HTML plots of the results.
Finds HVAC loops, returns their topology and draws loop diagrams.

Limitations

Version 0.1.0. The supported setup builds a local Docker image that bundles EnergyPlus.
Works on IDF files only; OpenStudio OSM models need the OpenStudio server instead.
HTTP mode needs bearer tokens you generate and configure yourself.

Install

git clone https://github.com/LBNL-ETA/EnergyPlus-MCP.git
cd EnergyPlus-MCP
docker build -t energyplus-mcp-dev -f .devcontainer/Dockerfile .devcontainer

Source: github.com/LBNL-ETA/EnergyPlus-MCP#claude-desktop

{
  "mcpServers": {
    "energyplus": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-v", "/path/to/EnergyPlus-MCP:/workspace",
        "-w", "/workspace/energyplus-mcp-server",
        "energyplus-mcp-dev",
        "uv", "run", "python", "-m", "energyplus_mcp_server.server"
      ]
    }
  }
}

Source: github.com/LBNL-ETA/EnergyPlus-MCP#claude-desktop

MCP servers and skills run code on your computer. Review the source before installing.

AI-generated models, drawings and results must be checked by a qualified engineer before use.