Adios MCP Server

Part of CLIO Kit - Gnosis Research Center

ADIOS MCP is a comprehensive Model Context Protocol (MCP) server that enables Language Learning Models (LLMs) to access and analyze scientific simulation and real-time data through the ADIOS2 framework. This server provides read-only access to BP5 datasets with intelligent data handling and seaml...

Quick Start

uvx clio-kit mcp-server adios

Documentation

• Full Documentation: CLIO Kit Website
• Installation Guide: See INSTALLATION.md
• Contributing: See Contribution Guide

Description

ADIOS MCP is a comprehensive Model Context Protocol (MCP) server that enables Language Learning Models (LLMs) to access and analyze scientific simulation and real-time data through the ADIOS2 framework. This server provides read-only access to BP5 datasets with intelligent data handling and seamless integration with AI coding assistants.

Key Features:

• High-Performance Data Access: Read-only access to ADIOS2 BP5 files with optimized I/O operations
• Intelligent Data Processing: Automatic variable type detection, shape analysis, and step-based data handling
• Multiple Data Operations: Variable inspection, attribute reading, data slicing, and statistical analysis
• Scientific Computing Support: Time-series analysis, multi-dimensional arrays, and metadata handling
• Cross-Platform Compatibility: Pure Python implementation for Linux, macOS, and Windows
• MCP Integration: Full Model Context Protocol compliance for seamless LLM integration

🛠️ Installation

Requirements

• Python 3.10 or higher
• uv package manager (recommended)
• Linux/macOS environment (Windows supported)

{
  "mcpServers": {
    "adios-mcp": {
      "command": "uvx",
      "args": ["clio-kit", "mcp-server", "adios"]
    }
  }
}

"mcp": {
  "servers": {
    "adios-mcp": {
      "type": "stdio",
      "command": "uvx",
      "args": ["clio-kit", "mcp-server", "adios"]
    }
  }
}

claude mcp add adios-mcp -- uvx clio-kit mcp-server adios

{
  "mcpServers": {
    "adios-mcp": {
      "command": "uvx",
      "args": ["clio-kit", "mcp-server", "adios"]
    }
  }
}

CLONE_DIR=$(pwd)
git clone https://github.com/iowarp/clio-kit.git
uv --directory=$CLONE_DIR/clio-kit/clio-kit-mcp-servers/adios run adios-mcp --help

set CLONE_DIR=%cd%
git clone https://github.com/iowarp/clio-kit.git
uv --directory=%CLONE_DIR%\clio-kit\clio-kit-mcp-servers\adios run adios-mcp --help

$env:CLONE_DIR=$PWD
git clone https://github.com/iowarp/clio-kit.git
uv --directory=$env:CLONE_DIR\clio-kit\clio-kit-mcp-servers\adios run adios-mcp --help

Capabilities

list_bp5

Description: Lists all BP5 files in a given directory. The 'directory' parameter must be an absolute path. Hints: read-only, idempotent Tags: adios, scientific-io

inspect_variables

Description: Inspects variables in a BP5 file, returning type, shape, and steps. Optionally filters by variable name. Hints: read-only, idempotent Tags: adios, variables

inspect_variables_at_step

Description: Inspects a specific variable at a given step in a BP5 file. All parameters are required. Hints: read-only, idempotent Tags: adios, variables

inspect_attributes

Description: Reads global or variable-specific attributes from a BP5 file. The 'variable_name' is optional. Hints: read-only, idempotent Tags: adios, attributes

read_variable_at_step

Description: Reads a named variable at a specific step from a BP5 file. All parameters are required. Hints: read-only, idempotent Tags: adios, variables

Resources

• adios://capabilities - ADIOS2 file format capabilities and supported engines.

Prompts

• explore_bp_file: Guided workflow for exploring an ADIOS2 BP file.

Claude Code

claude mcp add clio-adios -- uvx clio-kit adios

Or install via the CLIO Kit plugin marketplace:

/plugin marketplace add iowarp/clio-kit
/plugin install clio-adios@iowarp-clio-kit

Claude Desktop

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "clio-adios": {
      "command": "uvx",
      "args": [
        "clio-kit",
        "adios"
      ]
    }
  }
}

Gemini CLI

Add to ~/.gemini/settings.json:

{
  "mcpServers": {
    "clio-adios": {
      "command": "uvx",
      "args": [
        "clio-kit",
        "adios"
      ]
    }
  }
}

Or install the CLIO Kit extension:

gemini extensions install https://github.com/iowarp/clio-kit

Examples

1. Scientific Data Structure Analysis

I have a BP5 simulation file at /data/simulation_results.bp. Can you first analyze the data structure and then show me the temperature variable evolution over time?

Tools called:

• inspect_variables - Analyze the dataset structure and available variables
• read_variable_at_step - Read temperature data at specific time steps

This prompt will:

• Use inspect_variables to analyze the BP5 file structure and discover all variables
• Extract temperature variable data using read_variable_at_step at different time steps
• Provide insights about the simulation data structure and temporal evolution

2. Multi-Variable Scientific Analysis

Using the file /data/fluid_dynamics.bp, perform a comprehensive analysis showing:
1. All available variables and their properties
2. Pressure field at step 50
3. Variable metadata and attributes for the pressure field

Tools called:

• inspect_variables - List all available variables and properties
• read_variable_at_step - Extract pressure field at step 50
• inspect_attributes - Get pressure variable metadata

This prompt will:

• Generate comprehensive variable inventory using inspect_variables
• Extract specific time step data using read_variable_at_step
• Analyze variable metadata with inspect_attributes
• Provide detailed analysis of fluid dynamics simulation data

3. Time-Series Variable Inspection

From /data/climate_model.bp, I want to understand the temperature variable structure across different time steps and inspect its properties at step 25.

Tools called:

• inspect_variables - Get temperature variable structure and available steps
• inspect_variables_at_step - Detailed inspection at step 25

This prompt will:

• Use inspect_variables to understand overall variable structure
• Use inspect_variables_at_step to get detailed information at a specific time step
• Provide comprehensive time-series data structure analysis

4. High-Performance Computing Data Analysis

Analyze the simulation output at /data/parallel_computation.bp - show me the variable structure, read specific computational domain data, and extract metadata attributes.

Tools called:

• inspect_variables - Analyze computational variables and structure
• inspect_attributes - Extract simulation metadata and parameters
• read_variable_at_step - Read domain-specific data

This prompt will:

• Use inspect_variables to understand parallel computation structure
• Extract simulation parameters with inspect_attributes
• Read specific computational domain data using read_variable_at_step
• Provide HPC-focused data analysis

5. Scientific Data Validation and Exploration

I need to validate the integrity of my ADIOS dataset at /data/experimental_results.bp - check all variables, their step-wise properties, and ensure metadata completeness.

Tools called:

• inspect_variables - Comprehensive variable structure validation
• inspect_variables_at_step - Step-specific variable validation
• inspect_attributes - Metadata integrity check

This prompt will:

• Use inspect_variables to validate overall data structure integrity
• Perform step-specific analysis with inspect_variables_at_step
• Check metadata completeness with inspect_attributes
• Provide comprehensive data quality assessment for scientific workflows

6. Quick BP5 File Discovery

Show me all BP5 files in my simulation directory at /data/simulations/ and provide a quick overview of their contents and structure.

Tools called:

• list_bp5 - Directory-wide BP5 file discovery
• inspect_variables - Quick structure overview for each file

This prompt will:

• Use list_bp5 to discover all BP5 files in the directory
• Use inspect_variables to provide structural overview of discovered files
• Generate comprehensive overview of available simulation datasets
• Suggest optimal data access strategies based on file contents