VeduisWeb/mcp-server-directory

Model Context Protocol (MCP) Server Directory

Curated, schema-validated dataset of 800 Model Context Protocol (MCP) servers, category taxonomies, execution commands, and repository metadata for AI agents and tool-augmented LLM systems.

Curated and maintained by VePrompts and Veduis.

---

Dataset Summary

The Model Context Protocol (MCP), open-sourced by Anthropic in late 2024, standardized how LLMs connect to external development tools, databases, APIs, and client runtimes.

This dataset aggregates 800 active, community-verified and official MCP servers across 12 primary domains, including:

• Databases: PostgreSQL, MySQL, SQLite, Redis, MongoDB, Supabase.
• Developer Tools: GitHub, GitLab, Docker, Kubernetes, Sentry, Jira.
• Cloud & Infrastructure: Cloudflare, AWS, Heroku, DigitalOcean.
• Search & Research: Web search, Brave Search, ArXiv, Wikipedia.
• Productivity & Communication: Slack, Discord, Google Drive, Notion.

The dataset powers the live public directory at veprompts.com/mcp-servers/ and is archived on Zenodo (CERN).

---

Supported Tasks and Use Cases

• Agent Tool Discovery: Querying relevant MCP servers based on user query intent.
• Schema Validation: Verifying server configuration parameters and transport protocols.
• Ecosystem Benchmarking: Tracking implementation language distribution, star ratings, and maintenance activity across the MCP open-source ecosystem.

---

Data Fields

Field	Type	Description
slug	string	Unique URL slug and identifier
name	string	Normalized package or server name
title	string	Display title
author	string	Author or maintaining organization
author_url	string	URL of author profile or homepage
repo_url	string	GitHub or git repository URL
npm_package	string / null	Published npm package identifier if available
pypi_package	string / null	Published PyPI package identifier if available
category	string	Primary functional category
subcategory	string	Secondary classification
tags	list[string]	Search and topic tags
language	string	Primary implementation language (TypeScript, Python, Go, Rust)
license	string	Open source license (MIT, Apache-2.0, etc.)
stars	int	GitHub star count at time of cataloging
official	bool	Whether the server is maintained by the upstream platform vendor
summary	string	Functional description of available tools and resources

---

Direct Usage

Python with Pandas

import pandas as pd

df = pd.read_csv("https://huggingface.co/datasets/VeduisWeb/mcp-server-directory/raw/main/mcp-servers.csv")
print(f"Loaded {len(df)} MCP servers")

# Filter for database servers
db_servers = df[df["category"] == "databases"]
print(db_servers[["name", "repo_url", "stars"]].head())

Python Package (CLI Discovery)

Install the zero-dependency CLI tool directly from PyPI:

pip install veprompts-mcp
veprompts-mcp search postgres

---

Research Citations & Related Projects

• VePrompts Directory: https://veprompts.com/mcp-servers/
• VePrompts Platform: https://veprompts.com/
• Veduis Core: https://veduis.com/
• vComplianceScanner Source Auditing: https://veduis.com/products/vcompliance-scanner/
• Zenodo CERN DOI: 10.5281/zenodo.23840519
• PyPI Package: https://pypi.org/project/veprompts-mcp/