An open-source MCP (Model Context Protocol) server that connects Claude directly to your Energy Star Portfolio Manager account. Explore what's possible when AI meets sustainability data in plain language instead of exporting spreadsheets.
Built by Nikolas Mirando as an independent prototype exploring AI + sustainability data infrastructure in CRE.
Once connected, you can ask things like:
- "What is the average ENERGY STAR score for my office properties?"
- "Which of my properties are below a score of 50?"
- "Show me energy use intensity across my multifamily assets, ranked worst to best."
- "How many of my properties have ENERGY STAR certification?"
- "Give me a portfolio summary — scores and EUI across all my properties."
Your credentials stay on your machine. The MCP server runs locally, calls the ESPM API using your credentials, and returns your data to Claude. Nothing touches any third-party server. For demonstration and educational purposes only. This is not an official ENERGY STAR, EPA, or employer-sponsored system. Users authenticate with their own ENERGY STAR credentials locally; no data is stored or transmitted by this project.
Claude (your question)
→ MCP server (runs locally on your machine)
→ ESPM API (using your credentials from .env)
→ Your portfolio data only
→ Claude answers your question
- Node.js v18 or higher
- Claude Desktop
- An Energy Star Portfolio Manager account with API access enabled (see below)
You need to enable web services on your ESPM account before the API will work.
For the test environment (try it immediately with sample data):
- Log in to the ESPM Test environment
- Go to Account Settings → Software Development tab
- Select "Yes" for "Will this account be used to test web services?"
- Click Start Using the Test Environment
For your live portfolio (requires EPA approval):
- Log in to Portfolio Manager
- Go to Account Settings → Software Development tab
- Fill out the registration form to request live API access
- EPA will review and approve — typically within a few business days
git clone https://github.com/nikmirando1/ESPM_MCP.git
cd ESPM_MCP
npm installcp env.example .envOpen .env and fill in your ESPM username and password:
ESPM_USERNAME=your_username
ESPM_PASSWORD=your_password
ESPM_ENV=test # change to "live" once you have live API access
Open your Claude Desktop config file:
- Mac:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add this entry (update the path to where you cloned the repo):
{
"mcpServers": {
"espm": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/ESPM_MCP/src/index.js"]
}
}
}Restart Claude Desktop. You'll see the ESPM tools available. Start asking questions about your portfolio.
| Tool | What it does |
|---|---|
get_account |
Confirm you're connected and see your account info |
list_properties |
List all property IDs in your account |
get_property |
Get details for a specific property |
get_property_metrics |
Get score, EUI, and GHG emissions for a property |
list_property_groups |
List all your property groups |
get_property_group |
Get details for a specific group |
get_group_score_summary |
Average score, min/max, and full breakdown for a group |
get_portfolio_summary |
Portfolio-wide summary across all properties |
get_energy_star_certification_summary |
Count properties certified in a specific year using ESPM certification metrics |
- This repo: Free
- ESPM API: Free
- Claude Desktop: Requires a Claude subscription (Pro or above)
- Hosting: None — runs entirely on your machine
Your ESPM credentials are stored only in the .env file on your local machine. They are never transmitted to any server other than the official ESPM API (portfoliomanager.energystar.gov). This MCP has no backend, no telemetry, and no external dependencies beyond the ESPM API itself.
As of early 2026, there is ongoing uncertainty around federal funding for the ENERGY STAR program. This MCP was partly built to illustrate a broader point: the AI tooling is ready, but the data access layer is often the bottleneck. Whether ESPM continues as-is or evolves, the pattern demonstrated here, connecting sustainability data to conversational AI, is one the CRE industry should be building toward.
PRs welcome. Particularly interested in:
- Additional metrics endpoints (water, waste, carbon)
- Support for sharing/connection workflows
- Better error handling for large portfolios
MIT