Codesys-mcp-toolkit
A Model Context Protocol (MCP) server for CODESYS V3 programming environments.
@codesys/mcp-toolkit
A Model Context Protocol (MCP) server for CODESYS V3 programming environments. This toolkit enables seamless interaction between MCP clients (like Claude Desktop) and CODESYS, allowing automation of project management, POU creation, code editing, and compilation tasks via the CODESYS Scripting Engine.
🌟 Features
-
Project Management
- Open existing CODESYS projects (
open_project) - Create new projects from standard templates (
create_project) - Save project changes (
save_project)
- Open existing CODESYS projects (
-
POU Management
- Create Programs, Function Blocks, and Functions (
create_pou) - Set declaration and implementation code (
set_pou_code) - Create properties for Function Blocks (
create_property) - Create methods for Function Blocks (
create_method) - Compile projects (
compile_project)
- Create Programs, Function Blocks, and Functions (
-
MCP Resources
codesys://project/status: Check scripting status and currently open project state.codesys://project/{+project_path}/structure: Retrieve the object structure of a specified project.codesys://project/{+project_path}/pou/{+pou_path}/code: Read the declaration and implementation code for a specified POU, Method, or Property accessor.
📋 Prerequisites
- CODESYS V3: A working CODESYS V3 installation (tested with 3.5 SP21) with the Scripting Engine component enabled during installation.
- Node.js: Version 18.0.0 or later is recommended.
- MCP Client: An MCP-enabled application (e.g., Claude Desktop).
(Note: CODESYS uses Python 2.7 internally for its scripting engine, but this toolkit handles the interaction; you do not need to manage Python separately.)
🚀 Installation
The recommended way to install is globally using npm:
npm install -g @codesys/mcp-toolkit
This installs the package globally, making the codesys-mcp-tool command available in your system's terminal PATH.
(Advanced users can also install from source for development - see CONTRIBUTING.md if available).
🔧 Configuration (IMPORTANT!)
This toolkit needs to know where your CODESYS installation is and which profile to use. Configuration is typically done within your MCP Client application (like Claude Desktop).
Recommended Configuration Method (Direct Command)
Due to potential environment variable issues (especially with PATH) when launching Node.js tools via wrappers like npx within certain host applications (e.g., Claude Desktop), it is strongly recommended to configure your MCP client to run the installed command codesys-mcp-tool directly.
Example for Claude Desktop (settings.json -> mcpServers):
{
"mcpServers": {
// ... other servers ...
"codesys_local": {
"command": "codesys-mcp-tool", // <<< Use the direct command name
"args": [
// Pass arguments directly to the tool using flags
"--codesys-path", "C:\\Program Files\\Path\\To\\Your\\CODESYS\\Common\\CODESYS.exe",
"--codesys-profile", "Your CODESYS Profile Name"
// Optional: Add --workspace "/path/to/your/projects" if needed
]
}
// ... other servers ...
}
}
Key Steps:
- Replace
"C:\\Program Files\\Path\\To\\Your\\CODESYS\\Common\\CODESYS.exe"with the full, correct path to your specificCODESYS.exefile. - Replace
"Your CODESYS Profile Name"with the exact name of the CODESYS profile you want to use (visible in the CODESYS UI). - Ensure the
codesys-mcp-toolcommand is accessible in the system PATH where the MCP Client application runs. Global installation vianpm install -gusually handles this. - Restart your MCP Client application (e.g., Claude Desktop) to apply the settings changes.
Alternative Configuration (Using npx - Not Recommended)
Launching with npx has been observed to cause immediate errors ('C:\Program' is not recognized...) in some environments, likely due to how npx handles the execution environment. Use the Direct Command method above if possible. If you must use npx:
// Example using npx (POTENTIALLY PROBLEMATIC - USE WITH CAUTION):
{
"mcpServers": {
"codesys_local": {
"command": "npx",
"args": [
"-y", // Tells npx to install temporarily if not found globally
"@codesys/mcp-toolkit",
// Arguments for the tool MUST come AFTER the package name
"--codesys-path", "C:\\Program Files\\Path\\To\\Your\\CODESYS\\Common\\CODESYS.exe",
"--codesys-profile", "Your CODESYS Profile Name"
]
}
}
}
(Note: The -- separator after the package name might sometimes help npx but is not guaranteed to fix the environment issue.)
🛠️ Command-Line Arguments
When running codesys-mcp-tool directly or configuring it, you can use these arguments:
-p, --codesys-path <path>: Full path toCODESYS.exe. (Required, overridesCODESYS_PATHenv var, has a default but relying on it is not recommended).-f, --codesys-profile <profile>: Name of the CODESYS profile. (Required, overridesCODESYS_PROFILEenv var, has a default but relying on it is not recommended).-w, --workspace <dir>: Workspace directory for resolving relative project paths passed to tools. Defaults to the directory where the command was launched (which might be unpredictable when run by another application). Setting this explicitly might be needed if using relative paths.-h, --help: Show help message.--version: Show package version.
🔍 Troubleshooting
-
'C:\Program' is not recognized...error immediately after connection:- Cause: This typically happens when the tool is launched via
npxwithin an environment like Claude Desktop. The execution environment (PATHvariable) provided to the process likely causes an internal CODESYS command (like running Python) to fail. - Solution: Configure your MCP Client to run the command directly (
"command": "codesys-mcp-tool") instead of using"command": "npx". See the Recommended Configuration Method section above.
- Cause: This typically happens when the tool is launched via
-
Tool Fails / Errors in Output:
- Check the logs from your MCP Client application (e.g., Claude Desktop logs). Look for
INTEROP:messages or PythonDEBUG:/ERROR:messages printed to stderr from the CODESYS script execution. - Ensure the
--codesys-pathand--codesys-profilearguments passed to the command are correct and point to a valid CODESYS installation with scripting enabled. - Verify the project paths and object paths you are passing to tools are correct (use forward slashes
/). - Make sure no other CODESYS instances are running in conflicting ways (e.g., holding a lock on the profile).
- Check the logs from your MCP Client application (e.g., Claude Desktop logs). Look for
-
command not found: codesys-mcp-tool:- Ensure the package was installed globally (
npm install -g @codesys/mcp-toolkit). - Ensure the npm global bin directory is in your system's
PATHenvironment variable. Find it withnpm config get prefixand add thebinsubdirectory (or the main directory itself on Windows) to your PATH.
- Ensure the package was installed globally (
-
Check Logs:
- Claude Desktop logs:
C:\Users\<YourUsername>\AppData\Roaming\Claude\logs\(Windows)
- Claude Desktop logs:
🤝 Contributing
Contributions, issues, and feature requests are welcome! Feel free to check the issues page. (Optionally add a CONTRIBUTING.md file with more details).
📝 License
This project is licensed under the MIT License - see the LICENSE file for details.
🙏 Acknowledgements
- The CODESYS GmbH team for the powerful CODESYS platform and its scripting engine.
- The Model Context Protocol project for defining the interaction standard.
- All contributors and users who help improve this toolkit.
Related Servers
HAL (HTTP API Layer)
An MCP server that enables Large Language Models to make HTTP requests and interact with web APIs. It supports automatic tool generation from OpenAPI/Swagger specifications.
Chronulus AI
Predict anything with Chronulus AI forecasting and prediction agents.
xMCP Server
A streamable HTTP MCP server that proxies requests to stdio MCP servers within a container, providing a consistent command environment.
Unity-MCP
A bridge between the Unity game engine and AI assistants using the Model Context Protocol (MCP).
MCP SBOM Server
Performs a Trivy scan to produce a Software Bill of Materials (SBOM) in CycloneDX format.
MCP Sandbox
An interactive sandbox to safely execute Python code and install packages in isolated Docker containers.
Authless Remote MCP Server
An example of a remote MCP server deployable on Cloudflare Workers without authentication.
MCP Talk Demo Files
A collection of demo files for MCP servers and clients, illustrating various transport protocols and server capabilities using Python.
Galley MCP Server
Integrates Galley's GraphQL API with MCP clients. It automatically introspects the GraphQL schema for seamless use with tools like Claude and VS Code.
Sherlog MCP Server
A persistent IPython workspace for data analysis, log processing, and multi-agent collaboration.