मुख्य सामग्री पर जाएं

skills · The Model Context Protocol

MCP (Model Context Protocol)

An MCP server on your own data, tested and limited.

कोर्स के अंत तक, आप ये कर पाएँगे:

  • Protocol messages you can read
  • Tool descriptions written for a model to choose from
  • Checked inputs, and errors that say what to do next
  • Resources and prompts
  • Tests that call the server through a client session
  • A permission table, a planted-text test and a write that needs a preview
  • A README another person can install from

$9/माह से7.5 घंटे5 अध्याय20 पाठ

कोर्स अभी अंग्रेज़ी में हैं।

“Your server's caller is a model.”

इस कोर्स के बारे में

यह किसके लिए है

  • You write basic Python and want an AI assistant to use a tool or data source you work with
  • You have heard of MCP and want to read what the protocol sends between a client and a server
  • You want a server with limits clear enough that you would let someone else install it

आपको क्या चाहिए

  • Basic Python and a terminal
  • Node, for the MCP Inspector

आप कौन से कौशल अभ्यास करेंगे

  • Model Context Protocol
  • MCP servers
  • JSON-RPC
  • Tool design
  • Input validation
  • Client-session testing
  • Least privilege
  • Prompt injection testing

कितना समय लगता है

कुल मिलाकर लगभग 7.5 घंटे

  1. अध्याय 1 · Why a protocol1.5 घंटे
  2. अध्याय 2 · Tools a model can use1.5 घंटे
  3. अध्याय 3 · Resources, prompts, and tests1.5 घंटे
  4. अध्याय 4 · What it must never do1.5 घंटे
  5. अध्याय 5 · Ship it1.5 घंटे

इसमें क्या शामिल है

  • 20 पाठ
  • 77 पठन
  • 20 हल किए गए उदाहरण
  • 40 अभ्यास
  • 40 त्वरित जाँचें
  • 20 आम भ्रम
  • 20 क्विज़
  • अध्याय के अंत की 5 जाँचें

पाठ्यक्रम

5 अध्याय, 20 पाठ

पूरा कोर्स: हर अध्याय और पाठ। विस्तार के लिए अध्याय पर टैप करें। शुरू करने पर पाठ खुलेंगे।

    • The Description Is a Prompt
    • Check Every Input
    • Errors the Model Can Act On
    • Fewer, Task-Shaped Tools
    • Resources: Data the App Chooses
    • Resource Templates
    • Prompts the User Picks
    • Test It Like a Client
    • Tool Output Is Untrusted Input
    • Least Privilege
    • Writes Need a Person
    • Secrets and Where the Server Runs
    • Going Remote: Streamable HTTP
    • Logs You Can Read Later
    • The Protocol Moves
    • The README Someone Else Can Install

करियर

यह कोर्स आपको कहाँ तक ले जा सकता है

वे भूमिकाएँ जिनसे यह कोर्स जुड़ता है, और वे कौशल जिन्हें कोर्स पूरा करने के बाद आप अपने रिज़्यूमे और LinkedIn पर लिख सकते हैं।

वे कौशल जो आप बनाएँगे

कोर्स पूरा करने के बाद इन्हें अपने रिज़्यूमे या LinkedIn प्रोफ़ाइल में चिपकाएँ।

  • Scoping an MCP server for one data sourceExplain what the Model Context Protocol standardizes and what it leaves to the server author
  • Reading MCP protocol messagesRead and debug JSON-RPC 2.0 requests, results and errors
  • Building an MCP server tool in PythonImplement MCP tools with the official Python SDK
  • Connecting an MCP server to a host over stdioConfigure a local MCP server in an AI host application
  • Writing MCP tool descriptions for a modelWrite tool names, descriptions and parameter documentation for LLM tool use
  • Validating tool inputs on an MCP serverValidate tool arguments with schema constraints and code checks
  • Designing tool errors a model can act onReturn tool execution errors with isError and an actionable message
  • Shaping an MCP tool list around user tasksDesign task-oriented tools instead of one tool per table or endpoint
  • Choosing between MCP tools, resources and promptsDesign MCP servers that expose tools, resources and prompts appropriately
  • Building MCP resource templates with checked parametersImplement parameterized MCP resources with URI templates
  • Writing MCP prompts for user-started tasksAuthor reusable MCP prompt templates with arguments
  • Testing an MCP server through a client sessionWrite automated tests for MCP servers with pytest and the MCP Python SDK client
  • Treating tool output as untrusted inputReduce prompt injection risk in tool results returned to an AI host
  • Applying least privilege to an MCP serverScope an MCP server's data access with read-only connections and allow-lists
  • Designing write actions that wait for a personAnnotate MCP tools as read-only or destructive
  • Handling secrets and reach for an MCP serverKeep API keys in environment variables and out of tool results, code and version control
  • Running an MCP server over Streamable HTTP and deciding local or remoteRun MCP servers over stdio and Streamable HTTP transports
  • Structured logging for MCP tool callsAdd structured logging to MCP tool calls without logging secrets
  • Pinning SDK versions and tracking MCP protocol revisionsPin dependencies and track MCP protocol version changes
  • Documenting an MCP server's install steps, limits and threatsWrite installation and configuration documentation for MCP servers

प्रश्न

प्रतिबद्ध होने से पहले

You are studying or starting out in computing or engineering, you write basic Python, and you want to give an AI assistant tested, limited access to one tool or data source you use.

You do not need an API key to build and test the server, because the tests call it through the SDK's own client. From lesson 1.4 you also connect it to a host application such as Claude Desktop, which uses that application's account.

The course does not promise that, and no model was run to write it. You write descriptions for a model to choose from, then ask your own host five fixed questions and record which tool it called for each.

No. The lessons show the setup for Claude Desktop and the command for Claude Code. Another host can start the same server if it supports the stdio transport and a protocol version the server supports.

Your own source makes the course worth more, and each lesson has one required step on it. If you have none you may share, take the invented sample database, which has a path through each lesson.

आज ही शुरू करें