Model Context Protocol
Model Context Protocol (MCP) explained: what it is, why it exists, how it works, and the host/client/server architecture.
Public authority for MCP server discovery
Public authority for Model Context Protocol discovery. Every server and every claim is backed by a primary-source evidence record.
Every entry on this site references a primary source — a registry query, a package listing, a specification clause, or a verified repository. No marketing claims, no hosting, no pricing.
82 published, indexable editorial nodes across protocol fundamentals, guides, security, glossary, and reference pillars.
Start with Learn MCP →The Model Context Protocol defines how clients and servers exchange tools, resources, and prompts over JSON-RPC. Explore the protocol primitives and transport choices below.
Model Context Protocol (MCP) explained: what it is, why it exists, how it works, and the host/client/server architecture.
What an MCP server is, what it exposes, how local and remote servers differ, and how to choose and run one.
What an MCP client is, the client lifecycle, tool and resource calls, and which applications function as clients.
Verified server entities are currently under evidence review. See the directory for the latest review status.
Claude Desktop MCP support: local and remote servers, configuration, and how to add an MCP server.
Claude Code MCP support: adding servers to the coding agent and accessing tools from the terminal.
Cursor MCP support: stdio and Streamable HTTP transports, configuration, and troubleshooting.
VS Code MCP support: managed servers, user vs workspace configuration, and local and remote servers.
GitHub Copilot MCP support and configuration for agent tools.
OpenAI API remote MCP tools, allowed_tools, and read_only filters.
OpenAI Codex MCP server configuration.
Windsurf MCP server configuration.
Gemini CLI MCP server configuration.
Running and connecting MCP servers with Docker: containerized servers and networking.
How to build an MCP server: architecture, primitives, transport, authentication, testing, and publishing.
How to build an MCP client: host integration, lifecycle, primitive operations, and transport.
Build an MCP server with the official TypeScript SDK.
Build an MCP server with the official Python SDK.
MCP server metadata: what server.json is and how registry publication metadata is structured.
How to publish an MCP server: package, document, declare metadata, and submit to a registry.
How MCPserver.in selects and qualifies MCP servers, using documented capabilities rather than fabricated scores.
A decision framework for choosing an MCP server: maintainer, source, documentation, transport, permissions, and read/write behavior.
What verified status means at MCPserver.in: evidence, source provenance, and the difference between measurement and declaration.
How MCPserver.in compares MCP servers: what we compare, when comparisons are published, and the evidence requirements.
The difference between an MCP registry, a directory, a marketplace, and a discovery platform, and how to evaluate each.
How to evaluate MCP servers for databases, including read-only access and connection security.
Configuring a database MCP server in read-only mode to protect data from AI-agent calls.
MCP servers for developer workflows: version control, databases, and code tools.
MCP servers for browser automation: driving a browser from an AI agent and the security responsibility.
MCP servers for web search and research, and how they reduce reliance on stale context.
MCP servers for productivity tools: documents, communication, and task management.
MCP servers for infrastructure, deployment, cloud, and CI/CD, and the access risk.
MCP for enterprises: private registries, governance, audit logs, and multi-tenant controls.
Guide to Postgres MCP servers: what they do, how to evaluate them, read-only access, and connection security.
What Playwright MCP is, how it drives a browser, and the security responsibility.
Why an MCP server might not appear in Claude Desktop and how to fix it: config validation, paths, and restart.
Why Cursor MCP tools may not appear and how to fix it: configuration, reload, and transport.
How to diagnose MCP connection-refused and timeout errors across stdio and HTTP transports.
How to configure MCP servers in Claude Desktop: the config file, environment variables, and restart behavior.
How to configure MCP servers in VS Code at user and workspace scope, with examples.
How to secure and connect to remote MCP servers with OAuth, bearer tokens, and scopes.
How to evaluate GitHub MCP servers: read-only repository access, issues, and PR review.
An MCP security overview: threat model, authentication, tool permissions, prompt injection, and least privilege.
MCP authentication and authorization: bearer tokens, OAuth, scopes, and role-based access control.
MCP OAuth flow, protected resources, and how remote servers authorize access.
Prompt injection, tool poisoning, malicious servers, tool descriptions, and input validation in MCP.
Canonical definitions for MCP primitives, transports, and operational terminology.
Browse the glossary →How evidence is gathered, verified, and weighted. Every claim on this site is traceable to a primary source.