MCP Server Guide

What is an MCP Server?

An MCP server is a lightweight program that exposes tools, resources, and prompts to AI agents via the Model Context Protocol. Servers can provide access to databases, APIs, file systems, cloud services, and more.

When an AI agent connects to an MCP server, it can discover the server's capabilities and use them to complete tasks. For example, a PostgreSQL MCP server lets an AI agent query and manage databases.

Finding Servers

The VaultPlane registry organizes MCP servers by category:

• Databases — PostgreSQL, MySQL, SQLite, MongoDB, Redis
• Cloud & DevOps — AWS, GCP, Azure, Kubernetes, Docker
• CRM & Sales — Salesforce, HubSpot, Pipedrive
• Communication — Slack, Discord, Email, SMS
• Developer Tools — GitHub, GitLab, Jira, Linear
• Data & Analytics — BigQuery, Snowflake, Elasticsearch

Use the search bar to find servers by name or description, or filter by category.

Evaluating Servers

Before installing an MCP server, check these trust signals on the server detail page:

• Trust Score — A composite score (0-100) based on verification status, popularity, maintenance activity, and transparency.
• Verification Status — Unverified, Community, Verified, or Enterprise Verified.
• Permissions — What access the server needs and the risk level (low, medium, high, critical) of each permission.
• Compatibility — Which AI platforms the server supports (Claude Desktop, Cursor, LangChain, etc.).
• GitHub Stars & Downloads — Community adoption indicators.

Installing Servers

Most servers use npx for installation, which means you don't need to install anything globally:

For Claude Desktop, add the server to your configuration file:

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["@modelcontextprotocol/server-github"]
    }
  }
}

Each server's page on VaultPlane includes a copy-ready configuration snippet.

Next Steps