mcp-88 by alishams21 - MCP Server

MCP-88

MCP-88 is an MCP server designed for building data products. It does not provision data products directly, but instead supports their composition—covering aspects such as defining ownership, establishing data contracts, and enforcing policies. It enables users to create data products with a well-defined set of governance rules.

In short, it facilitates the development of data products within a data mesh architecture and promotes metadata-driven, "shift-left" governance—ensuring that clear governance steps are considered early in the data product lifecycle.

Status

Component	Status	Version	Description
Core MCP Server	🟢 Active	0.6.2	Main MCP server for data product composition
Scoping Agent	🟢 Active	0.6.2	Agent for data product scoping and requirements gathering
Data Contract Agent	🟢 Active	0.6.2	Agent for data contract definition and validation
Gradio Interface	🟢 Active	0.6.2	Web UI for interactive data product composition
Data Catalog Agent	⚪ Backlog	0.6.2	Agent for data catalog definition and validation
Deployment Agent	⚪ Backlog	0.6.2	Deployment composition of data products
Data Quality Port Agent	⚪ Backlog	0.6.2	Data quality port for data products
Data Observability Port Agent	⚪ Backlog	0.6.2	Data observability port for data products

Legend: 🟢 Active/Ready | 🟡 In Progress | ⚪ Backlog

Features

MCP server for composing data products
Agentic ai chat based approach for composing data products
Governance shift left in practice
Data mesh metadata shift left governance
Data product composition

How it works

MCP server is a server that provides a set of tools to the agent. The agent can use the tools to compose data products. The tools that current version of MCP-88 MCP server provides are:

Agent	Purpose	Description
Scoping Agent	Data Product Scoping	Captures product scope: name, domain, owner, purpose, and upstream data sources. Guides users through structured scoping process.
Data Contract Agent	Data Contract Definition	Defines data contracts with output ports, field schemas, sink locations, and freshness requirements. Handles field normalization and validation.
Data Catalog Agent	Data Catalog Management	Planned - Will manage data catalog entries, metadata, and lineage tracking for data products.
Deployment Agent	Data Product Deployment	Planned - Will handle deployment composition and orchestration of data products across different environments.
Data Quality Port Agent	Data Quality Management	Planned - Will define and manage data quality ports, validation rules, and quality metrics for data products.
Data Observability Port Agent	Data Observability	Planned - Will set up observability ports for monitoring, alerting, and tracking data product health and performance.

Quick Start

You can use MCP server either by normal way ot you can use a state-full agent with a MCP server which is developed on top of the MCP server.

set your environment variable:

OPENAI_API_KEY

Two example of uses of MCP server provided here so you can either simply run command line chat or gradio app, you can see structure in following:

make run-chat

├── chat/
│   └── chat.py          # ← Moved here with updated imports
├── demo/
│   ├── demo_server.py
│   ├── deploy_setup.py
│   ├── requirements-deploy.txt
│   └── start_demo_server.py
├── dp_chat_agent/
├── dp_composer_server/
├── Makefile             # ← Updated with new paths
└── ... (other files)

MCP Server Configuration

To run the dp_composer_server, you need to configure it as an MCP server in your client configuration. Add the following configuration to your MCP client settings (e.g., Claude Desktop configuration):

{
  "mcpServers": {
    "demo": {
      "command": "uv",
      "args": [
        "--directory",
        "<your-directory>/MCP-88/dp_composer_server",
        "run",
        "-m"
        "dp_composer-server"
      ]
    }
  }
}

This configuration allows the MCP client to connect to and use the data product composition tools provided by the MCP-88 server.