gilbert-mcp-server

ProduitsGilbert/gilbert-mcp-server

3.1

If you are the rightful owner of gilbert-mcp-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

The Gilbert-Tech MCP Server provides access to internal PLM data through Model Context Protocol tools, facilitating integration with Claude and other LLM applications.

Tools
5
Resources
0
Prompts
0

Gilbert-Tech MCP Server

A Model Context Protocol (MCP) server that provides access to Gilbert-Tech's internal PLM (Product Lifecycle Management) data. This server exposes PLM data through MCP tools that can be used by Claude and other LLM applications.

Features

  • Part Information Retrieval: Get detailed information about specific parts using part numbers
  • Search Functionality: Search for parts by name or description (planned)
  • Server Status: Check server health and configuration status
  • SSE Transport: Real-time communication using Server-Sent Events
  • Docker Support: Easy deployment with Docker and Docker Compose

Available Tools

PLM Tools

get_part_info(part_number: str)

Retrieve detailed information about a specific part from the PLM system.

Parameters:

  • part_number: The part number to look up

Returns:

  • Part details including specifications, status, and metadata
search_parts(search_term: str, limit: int = 10)

Search for parts in the PLM system (placeholder - implementation pending).

Parameters:

  • search_term: Term to search for
  • limit: Maximum number of results (default: 10)

Business Central ERP Tools

get_bc_entity_metadata(entity_name: str)

Discover available fields and structure for Business Central entities.

Parameters:

  • entity_name: BC entity name (e.g., 'Items', 'Vendors', 'PurchaseOrderHeaders')

Returns:

  • Entity metadata including available fields, endpoint information, and sample data structure

Common Entities:

  • Items: Product/inventory items
  • Vendors: Supplier information
  • Customers: Customer data
  • PurchaseOrderHeaders: Purchase order headers
  • PurchaseOrderLines: Purchase order line details
  • SalesOrderHeaders: Sales order headers
  • SalesOrderLines: Sales order line details
query_bc_entity(entity_name: str, filters: str = None, top: int = 10, select: str = None)

Query Business Central entities with OData filtering and field selection.

Parameters:

  • entity_name: BC entity to query
  • filters: OData filter expression (e.g., "No eq '12345'" or "Description contains 'motor'")
  • top: Maximum records to return (default: 10, max: 100)
  • select: Comma-separated fields to return (e.g., "No,Description,UnitPrice")

Filter Examples:

  • "No eq '12345'" - Exact item number match
  • "Description contains 'motor'" - Description contains text
  • "UnitPrice gt 100" - Price greater than 100
  • "CreatedDateTime ge 2024-01-01T00:00:00Z" - Created after date

SQL Server Database Tools

get_sql_database_schema(database_name: str)

Discover complete database schema including all tables, views, and statistics.

Parameters:

  • database_name: Database to query ('business_central' or 'cedule')

Returns:

  • Complete schema information organized by schemas
  • Table and view counts, database statistics
  • Server version and configuration details
get_sql_table_metadata(database_name: str, table_name: str, schema_name: str = "dbo")

Get detailed metadata for a specific table including columns, data types, and constraints.

Parameters:

  • database_name: Database to query ('business_central' or 'cedule')
  • table_name: Name of the table to analyze
  • schema_name: Schema name (default: 'dbo')

Returns:

  • Column definitions with data types, lengths, nullable status
  • Primary keys and constraints
  • Table statistics (row count, storage size)
sample_sql_table_data(database_name: str, table_name: str, schema_name: str = "dbo", limit: int = 10, columns: List[str] = None)

Get sample data from a table for analysis and development.

Parameters:

  • database_name: Database to query
  • table_name: Table to sample
  • schema_name: Schema name (default: 'dbo')
  • limit: Maximum rows (default: 10, max: 100)
  • columns: Specific columns to include (None = all)

Returns:

  • Sample data with actual values
  • Column names and data patterns
  • Data type examples
execute_sql_query(database_name: str, query: str, limit: int = 50)

Execute safe, read-only SQL queries with built-in security restrictions.

Parameters:

  • database_name: Database to query
  • query: SQL SELECT query to execute
  • limit: Maximum rows to return (default: 50, max: 100)

Returns:

  • Query results with column names
  • Row count and execution metadata

Safety Features:

  • Only SELECT statements allowed
  • Blocks dangerous keywords (DROP, DELETE, INSERT, UPDATE, etc.)
  • Automatic row limit enforcement
  • Read-only database connections
get_sql_database_status()

Get connection status and information for all configured SQL Server databases.

Returns:

  • Connection status for each database (business_central, cedule)
  • Server versions and database information
  • Configuration details and error messages

System Tools

get_server_status()

Get the current status of PLM, BC, and SQL server connections and configuration.

Deployment

Docker Compose (Recommended)

  1. Clone this repository to your Docker server (192.168.0.192)
  2. Ensure your PLM credentials are configured in data/config.ini
  3. Deploy using Docker Compose:
docker-compose up -d

The server will be available at: http://lpgadoc03:8002

Manual Docker

# Build the image
docker build -t gilbert-mcp-server .

# Run the container
docker run -d \
  --name gilbert-mcp-server \
  -p 8002:8002 \
  -v $(pwd)/data:/app/data:ro \
  gilbert-mcp-server

Configuration

PLM Credentials

Configure your PLM credentials using environment variables. Create a .env file:

# Copy the example file
cp .env.example .env

# Edit with your credentials
nano .env

Required environment variables:

PLM Configuration:

Business Central Configuration:

  • BC_username: Your BC username (format: DOMAIN\USERNAME)
  • BC_password: Your BC password or access token

SQL Server Database Configuration:

  • BC_SQL_server: Business Central SQL Server address
  • BC_SQL_database: Business Central database name
  • BC_SQL_username: SQL Server username for BC database
  • BC_SQL_password: SQL Server password for BC database
  • Cedule_SQL_server: Cedule CNC SQL Server address
  • Cedule_SQL_database: Cedule database name
  • Cedule_SQL_username: SQL Server username for Cedule database
  • Cedule_SQL_password: SQL Server password for Cedule database

Claude Desktop Integration

Method 1: JSON Configuration File (Recommended)

This method uses a configuration file and MCP proxy script for reliable connection.

Step 1: Download the MCP Proxy Script

Create the MCP proxy script that bridges HTTP API to MCP Protocol:

  1. Download the proxy script from this repository:

    curl -o mcp_proxy.py https://raw.githubusercontent.com/your-repo/gilbert-mcp-server/main/mcp_proxy.py

  2. Or create it manually by copying the content from mcp_proxy.py in this repository

  3. Make it executable:

    chmod +x mcp_proxy.py

  4. Update the server URL in the script if needed (line 22):

    BASE_URL = "http://lpgadoc03:8002"  # Change to your server address
Step 2: Create Claude Desktop Configuration

  1. Locate your Claude Desktop configuration directory:

    • macOS: ~/Library/Application Support/Claude/
    • Windows: %APPDATA%\Claude\

  2. Create or edit claude_desktop_config.json in that directory:

    {
  "mcpServers": {
    "gilbert-tech-plm": {
      "command": "python3",
      "args": [
        "/absolute/path/to/mcp_proxy.py"
      ]
    }
  }
}

  3. Replace the path with the absolute path to your mcp_proxy.py file

Step 3: Restart Claude Desktop
  1. Completely quit Claude Desktop
  2. Restart Claude Desktop
  3. Check for connection: You should see the Gilbert-Tech PLM tools available
Step 4: Verify Integration

Test the integration by asking Claude:

  • "What's the status of the PLM server?"
  • "Get information about part number [your-part-number]"
  • "What tools are available from Gilbert-Tech?"

Method 2: UI Integration (Limited Functionality)

This method uses Claude Desktop's "Add Integration" UI but has limitations.

Step 1: Open Claude Desktop Settings
  1. Open Claude Desktop
  2. Go to Settings (gear icon)
  3. Navigate to Integrations
  4. Click "Add Integration"
Step 2: Configure Integration
  1. Integration Name: gilbert-tech-plm
  2. Integration URL: http://lpgadoc03:8002
Step 3: Important Limitations

⚠️ Note: The UI method may have limited functionality because:

  • It typically expects SSE (Server-Sent Events) transport
  • Our server primarily uses HTTP transport
  • Some tools may not be properly exposed through this method

If the UI method doesn't work, use Method 1 (JSON Configuration) instead.

Troubleshooting

Claude Desktop Connection Issues

  1. Check server status:

    curl http://lpgadoc03:8002/health

  2. Verify tools are available:

    curl http://lpgadoc03:8002/mcp/tools/list

  3. Test a tool call:

    curl -X POST http://lpgadoc03:8002/mcp/tools/call \
  -H "Content-Type: application/json" \
  -d '{"name": "get_server_status", "arguments": {}}'

  4. Check Claude Desktop logs:

    • Look for MCP server logs in Claude Desktop
    • Check for error messages related to gilbert-tech-plm

  5. Verify proxy script:

    • Ensure Python 3 is available: python3 --version
    • Check dependencies: python3 -c "import mcp.server, httpx; print('Dependencies OK')"
    • Test proxy script: python3 mcp_proxy.py (should start without errors)

MCP Inspector for Advanced Troubleshooting

Use the MCP Inspector to test your proxy script:

npx @modelcontextprotocol/inspector python /path/to/mcp_proxy.py

This will open a web interface to test the MCP connection and tools.

Common Issues

Issue: "Server disconnected" or "Connection failed"

  • Solution: Ensure the proxy script path is absolute and correct
  • Solution: Check that the server URL in the proxy script matches your deployment

Issue: "Tools not showing in Claude"

  • Solution: Restart Claude Desktop completely
  • Solution: Check that the HTTP server is responding to tool requests

Issue: "Permission denied" errors

  • Solution: Make sure the proxy script is executable: chmod +x mcp_proxy.py
  • Solution: Verify Python 3 is in your PATH

Current Deployment

The Gilbert-Tech MCP server is currently deployed at:

  • Server URL: http://lpgadoc03:8002
  • Health Check: http://lpgadoc03:8002/health
  • Available Tools: 10 tools for PLM, Business Central, and SQL access
  • Transport: HTTP (SSE disabled)
  • Auto-restart: Enabled with Docker Compose

Claude Code (CLI)

Add this server to your Claude Code configuration:

Option 1: HTTP Transport
claude-code config servers add gilbert-tech-plm \
  --transport http \
  --url http://192.168.0.192:8002
Option 2: SSE Transport
claude-code config servers add gilbert-tech-plm \
  --transport sse \
  --url http://192.168.0.192:8002/sse
Configuration File Method

Add to your Claude Code configuration file (~/.config/claude-code/servers.json):

{
  "servers": {
    "gilbert-tech-plm": {
      "transport": "sse",
      "url": "http://192.168.0.192:8002/sse",
      "description": "Gilbert-Tech PLM data access"
    }
  }
}

Usage Examples

Once configured, you can use the MCP server in Claude:

What information do you have about part number "1510136"?

Get metadata for the Items entity in Business Central.

Query the first 5 items from Business Central where the description contains "motor".

Get purchase order headers created after January 1, 2024.

Get the database schema for the business_central database.

Show me the table structure for the Item table in business_central.

Sample 10 rows from the Vendor table in business_central.

Execute this query: SELECT TOP 5 * FROM dbo.Item WHERE Description LIKE '%motor%'

Check the status of all database connections (PLM, BC, and SQL).

API Endpoints

  • GET / - Server information and status
  • GET /health - Health check endpoint
  • GET /sse - SSE endpoint for MCP connections
  • POST /messages - Message handling endpoint
  • GET /docs - FastAPI documentation (Swagger UI)

Health Monitoring

The server includes health checks accessible at:

  • http://192.168.0.192:8002/health

Logs

View container logs:

docker-compose logs -f gilbert-mcp-server

Security Notes

  • The server runs on your internal network (192.168.0.192)
  • PLM credentials are stored in the mounted config file
  • CORS is enabled for development - consider restricting for production
  • Use HTTPS in production environments

Development

Local Development

# Install dependencies
pip install -r requirements.txt

# Run locally
python main.py

The server will start on http://localhost:8002

Testing

# Test the health endpoint
curl http://localhost:8002/health

# Test part lookup (replace with actual part number)
curl -X POST http://localhost:8002/messages \
  -H "Content-Type: application/json" \
  -d '{"method": "tools/call", "params": {"name": "get_part_info", "arguments": {"part_number": "TEST-123"}}}'

Troubleshooting

  1. Server won't start: Check Docker logs and ensure port 8002 is available
  2. PLM connection issues: Verify credentials in data/config.ini
  3. Claude can't connect: Ensure firewall allows access to port 8002 from your client machine
  4. SSL/TLS errors: PLM connection uses verify=False - ensure your PLM server certificate is valid

Version

Current version: 1.0.0