Your complete guide to MCP — the open standard for connecting AI applications to external systems
MCP is a standardized protocol that connects AI applications to any external system — like USB-C for your AI tools.
Expose tools for AI to execute actions, resources for context data, and prompts for interaction templates.
Built as an open-source standard with SDKs for TypeScript, Python, Go, Rust, and more.
The library runs to well over a hundred setup walkthroughs, so it is indexed twice. Browse guides by client and server type to narrow things down, or jump straight to one of the hubs below.
mcp.json config for every server we cover
Config paths, local files, troubleshooting
Postgres, warehouses and vector stores
AWS, containers and hosting platforms
If the protocol itself is new to you, our plain-English explainer on what MCP is covers the hosts, clients and servers involved, and the comparison of tools, resources and prompts explains why some servers can act on your behalf while others only hand over context. Before anything touches real credentials, read how MCP servers handle API keys, OAuth and service accounts and work through the server-side security checklist.
On the integration side, the most-read walkthroughs cover reading and transitioning Jira tickets from Cursor, managing a Linear backlog with a personal API key, creating and updating ClickUp tasks, and searching a Notion workspace from Composer. For infrastructure work there is querying S3, Lambda and CloudWatch through your AWS credential chain, running SQL against a BigQuery warehouse, and running MCP servers inside containers.
Rounding out the popular set: the Slack bot-token walkthrough that covers Claude Desktop and Cursor together, pulling Figma file structure into a design-to-code workflow, writing Cursor rules that fetch live library docs automatically, and the official GitHub server with its read-only mode and toolset flags.
How to connect a custom MCP server to ChatGPT in 2026: which plans get developer mode, where the Create / Add custom MCP server button lives, the remote-only rule, Secure MCP Tunnel for private servers, OAuth refresh-token gotchas, and how publishing freezes your tool list.
Configure MCP servers in Amazon Q Developer CLI: the global and workspace mcp.json file locations, the q mcp add/list/status commands, remote HTTP servers with OAuth, and troubleshooting servers that fail to load.
Connect Microsoft SQL Server to Cursor IDE with an MCP server so your AI can read schemas, write T-SQL, and debug queries without copy-pasting table structures. Covers the fragmented package landscape, a read-only login, and Azure SQL.
Connect Cursor to Buildkite with the official buildkite-mcp-server: Docker vs Homebrew install, BUILDKITE_API_TOKEN scopes, the OAuth-based remote server, and the compare_builds and list_jobs tools.
Connect Cursor to Pulumi Cloud via the official Remote MCP Server: HTTP transport config, OAuth setup, org selection, the get-stacks and resource-search tools, and Neo task delegation.
Connect Cursor to Jenkins with the official jenkinsci/mcp-server-plugin: install steps, the /mcp-server/mcp endpoint, Basic Auth with an API token, exposed tools, and troubleshooting.
The Model Context Protocol (MCP) is an open standard for connecting AI applications like Claude and ChatGPT to external systems such as databases, APIs, and file systems. Think of it as USB-C for AI — a standardized way for AI tools to interact with the outside world.
MCP was created by Anthropic and released as an open-source standard. It includes SDKs for multiple programming languages including TypeScript, Python, Go, and Rust.
With MCP, you can expose tools for AI to execute actions, provide resources for context data, and create prompt templates for interaction patterns. Common use cases include connecting AI to databases, integrating with APIs, accessing file systems, and building custom AI workflows.
Yes, MCP is completely free and open-source. You can use it in personal and commercial projects without any licensing fees.
MCP is supported by Claude (via Claude Desktop), and the ecosystem is growing. Many third-party tools and applications are adding MCP support, making it a versatile standard for AI integration.
New MCP server guides, setup tutorials, and integration tips — delivered weekly. No spam, unsubscribe anytime.