ALMA_MCP

ALMA MCP Server - Astronomical Data Access via Natural Language

A Model Context Protocol (MCP) server that provides comprehensive access to the ALMA (Atacama Large Millimeter/submillimeter Array) archive through a clean, extensible architecture.

This MCP server transforms ALMA archive queries from a software engineering problem into anatural language conversation. Instead of learning TAP/ADQL syntax and archive APIs, researchers simply ask for what they need and get clean, analysis-ready results.

The result: AI assistants that can seamlessly search ALMA data by target, position, frequency, resolution, or any custom criteria

IMPORTANT: Change the paths below according to YOUR installation location!

# Clone the repository (or copy the ALMA_MCP folder) git clone https://github.com/adamzacharia/ALMA_MCP.git cd ALMA_MCP # Create a dedicated conda environment with Python 3.11+ conda create -n alma_mcp python=3.11 conda activate alma_mcp # Install dependencies pip install -r requirements.txt # Install astronomical libraries for full functionality pip install fastmcp alminer pyvo astroquery astropy pandas

Alternative: Using venv instead of conda:

# Navigate to the ALMA_MCP folder cd path/to/ALMA_MCP # Create a dedicated venv environment python -m venv venv venv\Scripts\activate # Windows # source venv/bin/activate # Mac/Linux # Install dependencies pip install -r requirements.txt

# Test basic functionality python test_server.py # Quick test (optional) python -c " from server import get_alma_info, ALMINER_AVAILABLE, PYVO_AVAILABLE print(' Server loads successfully') print(f' alminer available: {ALMINER_AVAILABLE}') print(f' pyvo available: {PYVO_AVAILABLE}') "

Find and edit the Claude Desktop MCP configuration file:
-

Navigate to your Claude Desktop AppData folder:

- Windows: Open File Explorer and go to%APPDATA%\Claude\
- macOS:~/Library/Application Support/Claude/

add the 'claude_desktop_config.json' or add the below to your 'config.json' file

Add the following configuration to the file:

IMPORTANT: Change the path below according to YOUR installation location!

{ "mcpServers": { "alma": { "command": "python", "args": ["c:/Users/Asus/Desktop/Quasar-main/ALMA_MCP/server.py"] } } }

Note: Use forward slashes/in paths even on Windows.

Using a Custom Environment (Conda or venv)

If you installed the dependencies in aconda environmentorvenv, you MUST specify the full path to that environment's Python executable. Otherwise the system won't find the installed packages!

{ "mcpServers": { "alma": { "command": "C:/Users/YourName/anaconda3/envs/alma_mcp/python.exe", "args": ["c:/path/to/ALMA_MCP/server.py"], "cwd": "c:/path/to/ALMA_MCP", "env": {} } } }

To find your conda environment's Python path, run:

conda activate alma_mcp where python # Windows which python # Mac/Linux

{ "mcpServers": { "alma": { "command": "c:/path/to/ALMA_MCP/venv/Scripts/python.exe", "args": ["c:/path/to/ALMA_MCP/server.py"], "cwd": "c:/path/to/ALMA_MCP", "env": {} } } }

- Quit Claude Desktop completely(right-click system tray → Quit)
- Check Task Manager- If Claude is still running in the background, end the task
- Reopen Claude Desktop
- Test with a querylike:

- "Search ALMA for observations of M87"
- "Find high-resolution ALMA data with resolution under 0.5 arcseconds"
- "What are ALMA's frequency bands?"

# Check Python environment python --version # Should be 3.10+ # Test server manually python server.py # Should start without errors

- Verify the Python path in your config is correct
- Ensure all dependencies are installed
- Check thatserver.pyexists at the specified path

pip install fastmcp alminer pyvo astroquery astropy pandas

Once configured, you can ask natural language questions about ALMA data:

- "Find observations of Orion KL"
- "Search ALMA for M87 within 1 arcminute"
- "What ALMA data exists for NGC 1234?"

- "Search ALMA at RA=83.6, Dec=-5.4"
- "Find observations near the Galactic Center"
- "Cone search at coordinates 150.5, 2.2 with 5 arcmin radius"

- "Find ALMA data between 230 and 250 GHz"
- "Search for Band 6 observations"
- "What observations cover the CO(2-1) line at 230.5 GHz?"

- "Find high-resolution data with resolution under 0.1 arcseconds"
- "Search for ALMA observations better than 0.5 arcsec resolution"

- "Find ALMA proposals by Adele Plunket"
- "Get data for proposal 2023.1.00001.S"
- "Search for galaxy evolution proposals"

- "Do any M87 observations cover the CO(2-1) line?"
- "Check if Orion data covers HCN(1-0) at 88.6 GHz"
- "Find observations that cover 115 GHz for a source at z=0.5"

- "Run this SQL: SELECT target_name, band_list FROM ivoa.obscore WHERE target_name LIKE '%M87%'"
- "Query ALMA for all Band 7 observations from 2023"

- "What are ALMA's frequency bands?"
- "List common spectral lines in the mm range"
- "What science categories does ALMA support?"

- "Find ALMA observations published in Nature"
- "Search for data used in papers by Smith published in 2023"
- "What ALMA data has the bibcode 2017ApJ...834..140R"

- "Find all ALMA observations tagged with 'Quasars'"
- "Search for Sub-mm Galaxies (SMG) observations"
- "What observations are categorized under 'Active Galactic Nuclei'?"

- "Find all spectral cubes for M87"
- "Search for continuum images tagged with Exoplanets"
- "Show me spectral line observations in Band 6"

- "Find proposals mentioning black holes in their abstract"
- "Search for observations from proposals about star formation"
- "What proposals mention the cosmic microwave background?"

- "Find observations with continuum sensitivity better than 0.1 mJy/beam"
- "Search for deep ALMA observations under 0.05 mJy sensitivity"

- "Check if ALMA has observed M31, M51, M82, and NGC 1068"
- "Query ALMA for these sources: Cen A, M83, and NGC 1234"

You can use this MCP server with LangChain using thelangchain-mcp-adapterspackage:

from langchain_mcp_adapters.client import MCPClient from langchain_openai import ChatOpenAI # Connect to the MCP server client = MCPClient( command="python", args=["path/to/ALMA_MCP/server.py"] ) # Get tools from MCP server tools = client.get_tools() # Use with any LangChain-compatible LLM llm = ChatOpenAI(model="gpt-4") llm_with_tools = llm.bind_tools(tools)

For open source LLMs (Ollama, LMStudio, etc.), you can:
-

Use MCP-compatible clients: Some open source projects likeMCP CLIsupport connecting MCP servers to local LLMs.

from server import search_alma_by_target, search_alma_by_position, get_alma_info # Use tools directly result = search_alma_by_target("M87", search_radius_arcmin=5.0) print(result) # Get ALMA reference info info = get_alma_info() print(info)

ALMA_MCP/ ├── server.py # Main MCP server with 16 tools ├── requirements.txt # Python dependencies ├── test_server.py # Test suite └── README.md # This file

fastmcp>=2.0.0 # MCP framework pandas>=1.5.0 # Data manipulation

alminer>=0.2.0 # Advanced ALMA queries pyvo>=1.4.0 # TAP/ADQL access astroquery>=0.4.0 # SIMBAD, VizieR, etc. astropy>=5.0.0 # Astronomical utilities

@software{alma_mcp, title={ALMA MCP Server: Astronomical Data Access for AI Agents}, author={Adam Zacharia Anil and Adele Plunkett}, year={2025}, url={https://github.com/adamzacharia/ALMA_MCP} }

- Issues:GitHub Issues
- Documentation: This README and inline code comments
- Discussions:GitHub Discussions

An AI-powered API for medical research, unifying ClinicalTrials.gov, PubMed, and FDA databases with intelligent analysis.