MOF Tools MCP Server

A production-ready, professional MCP server for Metal-Organic Framework (MOF) research. This server uses the Streamable HTTP transport to provide scientific tools over HTTP with comprehensive Pydantic validation and a formal tool registry system.

✨ Key Features

• 🔒 Input/Output Validation: All tool inputs and outputs are validated using Pydantic models

• 📋 Formal Tool Registry: Centralized tool management with metadata, categories, and tags

• 🛡️ Production-Ready: Industry-standard code with comprehensive error handling

• 📊 Type Safety: Full type hints throughout the codebase

• ✅ Tested: Comprehensive test suite with over 35 tests

📁 Repository Structure

• main.py: Server entrypoint with tool registration and initialization

• tools.py: Core scientific tools with Pydantic validation models

• tool_registry.py: Formal tool registration system with metadata management

• tool_definitions.yaml: YAML configuration file defining all available tools and their metadata

• tests/test_tools.py: Comprehensive test suite for tools and validation

• tests/test_tools_load.py: Load and consistency tests for tool definitions

• pyproject.toml: Dependency and package management

🚀 Installation

🏃 Running the Server

To run as the modern HTTP server (Streamable HTTP):

The server will start on http://0.0.0.0:50001 and display registered tools information:

🧪 Testing

Run the comprehensive test suite:

🧪 Testing the Server

Method 1: MCP Inspector

Method 2: Manual HTTP Check

Since the server runs on HTTP, you can verify it's up with a simple curl:

🛠️ Available Tools

All tools return validated JSON responses with comprehensive error handling.

Bohr Agent SDK integration also provides asynchronous job management tools:

• submit_: Submit a calculation job

• query_job_status: Check job progress

• get_job_results: Retrieve completed job data

search_mofs

Category: Search
Description: Search for Metal-Organic Frameworks by name or formula in the database

Input:

Output:

calculate_energy

Category: Calculation
Description: Calculate the potential energy of a structure from CIF file content or path using ASE

Input:

Output:

optimize_structure

Category: Optimization
Description: Perform structure optimization for a named MOF structure (placeholder implementation)

Input:

Output:

🔌 Integration

To connect your agent to this server, use the Streamable HTTP endpoint:

URL: http://localhost:50001/mcp

🏗️ Architecture

Pydantic Validation

All tools use Pydantic models for comprehensive input/output validation:

• Input Models: Validate and sanitize user inputs

• Output Models: Ensure consistent response structure

• Field Validators: Custom validation logic (e.g., trimming whitespace, checking ranges)

• Error Handling: Graceful error messages returned as validated JSON

Tool Registry System

The formal tool registry provides:

• Metadata Management: Description, category, version, tags for each tool

• Categorization: Tools organized by category (search, calculation, optimization, analysis)

• Tag-based Discovery: Find tools by tags (e.g., "mof", "energy", "database")

Production-Ready Features

• ✅ Full type hints throughout

• ✅ Comprehensive error handling

• ✅ Input sanitization and validation

• ✅ Consistent JSON output format

• ✅ Detailed documentation

• ✅ Test coverage (38 tests)

• ✅ Modular, maintainable code structure

📚 Development

Tool Configuration

Tool definitions are stored in tool_definitions.yaml for easy configuration and management. This separates tool metadata from code logic.

Example tool definition:

Adding a New Tool

1. Define Pydantic models in tools.py:

2. Implement the tool function in tools.py:

3. Add the tool definition to tool_definitions.yaml:

The tool will be automatically registered when the server starts.

4. Add tests in test_tools.py:

📝 License

See LICENSE file for details.