eplusout-mcp
Provides comprehensive access to EnergyPlus building energy simulation results, enabling discovery, analysis, and extraction of data from epJSON, SQL, and HTML files through MCP tools with pandas integration.
README
EnergyPlus MCP Server
Overview
This Model Context Protocol (MCP) server provides comprehensive access to EnergyPlus building energy simulation results through a rich set of tools for discovering, analyzing, and extracting data from EnergyPlus model files. The server includes advanced features like pandas-based data analysis, keyword-based table searching, and comprehensive logging with token consumption tracking.
Key Features
- Comprehensive Data Access: Read epJSON input files, SQL result databases, and HTML summary reports
- Advanced Search Capabilities: Search HTML tables by keywords, find related epJSON objects, and explore model components
- Pandas Integration: Execute pandas queries directly on timeseries and tabular data
- Logging & Monitoring: Built-in token consumption tracking and function call monitoring
- Flexible Model Discovery: Automatic model cataloging and metadata extraction
Installation
# Clone the repository
git clone <repository-url>
cd mcp-eplus-outputs
# Install dependencies
uv sync
# Run the server
uv run main.py
Configuration
Add to your Claude Desktop configuration:
{
"mcpServers": {
"mcp_eplus_outputs": {
"command": "uv",
"args": ["--directory", "C:/path/to/mcp-eplus-outputs", "run", "main.py"]
}
}
}
File Structure
Each EnergyPlus model consists of three main file types:
.epJSON- Input model definition (building geometry, materials, HVAC systems, schedules).sql- Simulation results database (hourly timeseries data, summary tables).table.htm- HTML summary reports (tabular summaries of results)
Model Naming Convention
Files follow this pattern:
{CODENAME}_{PROTOTYPE}_{CODEYEAR}_{CITY}_{SKIPOPTIONS}_{HVAC_LABEL}.{EXTENSION}
Example: ASHRAE901_HotelLarge_STD2025_Buffalo_SkipEC_gshp.epJSON
Available Data
Building Types
- HotelLarge - Large hotel building prototype
- Warehouse - Warehouse building prototype
HVAC Systems
- gshp - Ground Source Heat Pump
- pkgdx_gas - Packaged DX with Gas
- pkgdx_hp - Packaged DX Heat Pump
- pvav_awhp - Packaged VAV with Air-to-Water Heat Pump
- pvav_blr - Packaged VAV with Boiler
- vav_ac_blr - VAV with Air-Cooled Chiller and Boiler
- vav_ac_blr_doas - VAV with Air-Cooled Chiller, Boiler, and DOAS
- vav_wc_blr - VAV with Water-Cooled Chiller and Boiler
- vrf - Variable Refrigerant Flow
- wshp_gas - Water Source Heat Pump with Gas
- pszvav_gas - Packaged Single Zone VAV with Gas
Locations
- Buffalo - Cold climate (upstate New York)
- Tampa - Hot climate (Florida)
Available Tools
Core Model Management
initialize_model_map()- Initialize model catalogget_available_models()- List all available models with metadataget_usage_instructions()- Get comprehensive usage documentation
HTML Table Analysis
get_html_table_by_tuple()- Retrieve specific HTML tablessearch_html_tables_by_keyword()- Find tables by keyword searchexecute_pandas_on_html_table()- Run pandas queries on HTML tablesexecute_multiline_pandas_on_html_table()- Run complex pandas code on HTML tables
Timeseries Data Analysis
get_sql_available_hourlies()- List available hourly variablesget_timeseries_report_by_rddid()- Extract timeseries data by RDD IDexecute_pandas_on_timeseries()- Run pandas queries on timeseries dataexecute_multiline_pandas_on_timeseries()- Run complex pandas code on timeseries data
epJSON Model Exploration
search_epjson_objects()- Search building model objectsget_object_properties()- Get detailed object propertieslist_objects_by_type()- List all objects of specific typesearch_related_objects()- Find related objects by pattern
General Data Processing
execute_query()- Execute pandas queries on cached dataexecute_multiline_query()- Execute multi-line pandas code on cached data
Quick Start Guide
1. Initialize the System
# Always start here
initialize_model_map(directory='eplus_files')
# Discover available models
models = get_available_models()
2. Explore Available Data
# Find cooling-related tables
cooling_tables = search_html_tables_by_keyword(
id='ASHRAE901|HotelLarge|STD2025|Buffalo|gshp',
keywords=['cooling', 'sizing', 'capacity']
)
# Get available timeseries variables
timeseries_vars = get_sql_available_hourlies(
id='ASHRAE901|HotelLarge|STD2025|Buffalo|gshp'
)
3. Extract and Analyze Data
# Get a specific HTML table
sizing_data = get_html_table_by_tuple(
id='ASHRAE901|HotelLarge|STD2025|Buffalo|gshp',
query_tuple=('Entire Facility', 'HVAC Sizing Summary', 'Zone Sensible Cooling')
)
# Analyze timeseries data with pandas
energy_analysis = execute_multiline_pandas_on_timeseries(
model_id='ASHRAE901|HotelLarge|STD2025|Buffalo|gshp',
rddid=179,
code='''
# Convert energy units and calculate monthly totals
df['kWh'] = df['Value'] / 3.6e6
df['month'] = df['dt'].dt.month
monthly_consumption = df.groupby('month')['kWh'].sum()
result = monthly_consumption.to_dict()
'''
)
Advanced Features
Pandas Integration
The server includes secure pandas execution environments for both HTML table and timeseries data:
- Single-line queries: Use
execute_pandas_on_*functions - Multi-line code: Use
execute_multiline_pandas_on_*functions withresult = ...pattern - Security: Restricted execution environment prevents dangerous operations
Keyword Search
Find relevant data using flexible keyword searching:
# Search for energy consumption tables
energy_tables = search_html_tables_by_keyword(
id=model_id,
keywords=['energy', 'consumption', 'end use'],
case_sensitive=False
)
Comprehensive Logging
All function calls are logged with token consumption tracking in monitor_logs/mcp_calls.log.
Performance Considerations
- Model map is cached for fast repeated access
- Large datasets are automatically truncated in responses
- HTML table search is optimized for performance
- Token consumption is monitored and logged
Error Handling
- Invalid model IDs return descriptive error messages
- Missing data returns empty results with status information
- Pandas execution errors are caught and reported safely
Token Management
The server includes comprehensive token counting and logging:
- Input/output tokens tracked per function call
- Logs stored in JSON format for analysis
- Automatic result truncation to prevent token overflow
Support
For detailed usage instructions and examples, use:
get_usage_instructions()
This returns the complete CLAUDE.md documentation file with comprehensive examples and best practices. "# eplusout-mcp"
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.