DMCPServer
Dinos MCP Server, make your code, on MCP Action and execute by AI
claude mcp add --transport stdio daniel09fernandes-dmcpserver docker run -i dmcpserver-image \ --env DMCP_LOGS="enable or disable logging for MCP requests" \ --env DMCP_HTTP_PORT="port for HTTP MCP endpoints" \ --env DMCP_STDIO_HOST="host for STDIO MCP connections"
How to use
DMCPServer is a Delphi-based Master Control Protocol server that enables exposing MCP actions over HTTP or STDIO. It lets you register actions (for example, get_weather) with input schemas and output handling, and then serve these actions to clients that send JSON-RPC-like requests. Typical usage involves installing the server, starting it, and then posting requests to its HTTP endpoint or communicating through STDIO streams, depending on your deployment. The README example shows how to define an action, parse input parameters, perform business logic via services, and return structured results or errors. You can also register server information, set versions, and attach tools or resources that your MCP can expose to clients or to memory-augmented workflows.
To use the available capabilities, you typically:
- Register actions with descriptive metadata and an input schema so clients know what to send.
- Implement the action logic using Delphi and your services, returning results in the MCP tooling format.
- Run the server in your preferred environment (for example, via Docker, as the README suggests).
- Expose the server via HTTP or STDIO so clients and other systems can invoke the actions and receive structured responses.
How to install
Prerequisites:
- Docker installed (or a native Delphi environment if you prefer running directly).
- Access to the boss tool if you want to install the server via the documented method.
Step-by-step installation:
-
Install Boss (if you plan to use boss install for package handling):
- Follow the Boss installation instructions from https://github.com/HashLoad/boss and ensure boss is in your PATH.
-
Install the DMCPServer from GitHub:
-
Using Boss (as per README):
boss install github.com/Daniel09Fernandes/DMCPServer
-
If you prefer a Docker-based approach (as reflected in the MCP config): pull or build an appropriate DMCPServer image following the project’s Docker guidance or Dockerfile (not provided in the snippet).
-
-
Run the server:
- If using Docker: ensure the image name matches the one in your config (dmcpserver-image) and start the container. You may need to map ports and volumes depending on whether you use HTTP or STDIO.
-
Verify:
- Access the HTTP endpoint as configured (e.g., http://localhost:<port>/) or connect via STDIO as per your environment.
-
Configuration:
- Adjust environment variables (DMCP_LOGS, DMCP_HTTP_PORT, DMCP_STDIO_HOST) to tailor logging and connection details to your deployment.
Additional notes
Tips and common issues:
- The DMCPServer in this repository uses Delphi-based tooling; if you deploy via Docker, ensure the image contains all runtime dependencies for a Delphi-based MCP server.
- If you see missing parameter errors, verify your action's input schema matches the expected fields (name, type, required).
- When using HTTP, ensure the port exposed by the container or host is accessible from clients. For STDIO, ensure your integration passes JSON objects correctly through the streams.
- Use the provided example get_weather action as a template to implement your own actions, including proper error handling and logging.
- If you modify server information (name, version, resources), keep consistency with your client-side expectations to avoid mismatches.
- The environment variable DMCP_HTTP_PORT is useful to avoid hard-coding ports in docker-compose or orchestration tooling.
Related MCP Servers
UE5
MCP for Unreal Engine 5
wanaku
Wanaku MCP Router
Delphi
Native Delphi Server implementation of the Model Context Protocol (MCP)
zerodha
Zerodha MCP Server & Client - AI Agent (w/Agno & w/Google ADK)
help-scout
MCP server for Help Scout - search conversations, threads, and inboxes with AI agents
mcp-manager
CLI tool for managing Model Context Protocol (MCP) servers in one place & using them across them different clients