MCP Desde Cero - Domina el Model Context Protocol HOY

Resources are data or content exposed by servers for clients, discoverable via lists or templates, with text or binary data, uri-based identifiers, supporting subscriptions and real-time updates via OAuth.

Explore prompts in MCP as templates for language models that support resources and multi-step flows, interface commands. Learn to list, fetch, and use prompts for code analysis and git commits.

Explain how tools in the model context protocol let servers perform calculations, interact with external APIs, and expose actions to clients, while guiding safe, authenticated usage.

Master sampling in MCP: enable servers to request LM responses (via clients) using messages with roles, system prompts, and token limits, processed under human supervision for safety and privacy.

Discover how routes in the MCP modal context protocol orient, clarify, and organize resources, and learn how clients declare, update, and validate routes with practical, real-world examples.

Explore MCP transports, including local stdio and http, and how JSON RPC messages—requests, responses, and notifications—traverse clients and servers. Learn about session IDs, streaming, endpoints, and security considerations.

discover community and official model context protocol servers, explore the repository and directory resources, and learn how to locate, evaluate, and install MCP servers for practical testing.

Install cloud desktop across Windows or Mac, note Linux lacks official support, and explore configuration, language settings, GitHub integration, and MCP server connections using the model context protocol.

Set up the first MCP server in cloud using the file system, grant a specific folder access to the AI agent, and configure, read, write, and manage files.

Install and configure an MCP server using cloud desktop or pip, then use the fetch tool to query internet URLs and return content as markdown, while debugging with logs.

Configure the MCP git server in cloud desktop to integrate git workflows, enabling commands like status, diff, and branch management directly in your repository.

Connect cloud to a database using supervise's MCP server with authentication tokens, base URL, and anon key. Create tables, run SQL queries, and validate an MCP MVP setup.

Build a minimal MCP server in TypeScript that greets users and exposes a tool list including grid tool, using npm and the modal context protocol SDK.

Master the MCP inspector to discover and test your MCP server, connect with a proxy session token, and run tools to list, call, and debug server actions.

Set up a hello world MCP server on Claude Desktop, run from the correct repository, and test agent greetings via cloud desktop.

Build an MCP server for a travel assistant that blends news, weather, and flight searches. Consume an external weather api to show current conditions and forecasts.

Initialize a fresh MCP project, install axios and other dependencies with npm, configure TypeScript, and create a basic structure with index and folders for tools, prompts, and resources.

Build a minimal MCP server from scratch, register tools like get weather, and fetch climate data from the OpenWeather API to return structured weather data for AI use.

Build a flight search tool on the MCP server using mock data to simulate an API, with from, to, date fields, and airlines like Aerolineas Argentinas, Latam, Jetsmart, Flybondi.

Create your first resource in the MCP server by adding airports data as a JSON resource, then learn to list and read resources from the server.

Integrate a server-side resource to search flights by deserializing airport data and matching from and to airports, returning a json response, and filter results by available airlines.

Define a city interface and populate a cities database to expose resources on the MVP server, including get cities resource and tourist info with attractions and travel tips for Argentina.

Add a distance calculator tool to the MVP server, computing distance between two cities with a simple latitude and longitude method, returning distance, time, and transport options.

Develop and deploy prompts for the MCP server, including a weather analysis prompt with city and days inputs, and expose prompts as independent resources for clients and internal use.

We complete all the project prompts with travel suggestions for destination, date, and preferences. Provide personalized recommendations on climate, luggage, activities by season, lodging, transport, and budget.

Create the final server tool for the MCP project that delivers travel recommendations by integrating city data, weather, tourism information, and prompts, with seasonal insights for Argentina.

The MCP client is the application-side consumer that requests data from an MCP server, enforcing authentication, authorization, and privacy to return only permitted data from sources like databases and APIs.

Exploro los clientes MCP disponibles y sus capacidades, incluyendo recursos, prompts y herramientas. Comparo ejemplos como cloud desktop, cursor y copilot para ilustrar los tipos de soporte.

Build a minimal MCP client to connect to a demo server and run a TypeScript project. Integrate a Gemini AI model via API, list tools, and enable a chat loop.