English(EN)What I Learned Building My First MCP Server
模型上下文协议 (MCP) 规范化AI工具交互
作者PulseAugur 编辑部·[41 个来源]·
模型上下文协议 (MCP) 正在成为一种标准化的方法，供AI应用程序发现和交互外部工具和资源，这与传统API不同。多篇文章探讨了MCP的架构、实现及其对AI代理开发的影响。关键方面包括它在使Claude和ChatGPT等AI模型能够访问搜索电子邮件或执行代码等功能方面的作用，MCP的AI到工具的关注点与API的软件到软件通信的区别，以及协议本身之外安全控制的重要性。实现范围从用于验证MCP端点健康的诊断工具到用于管理企业内部工具访问的完整网关。
AI
<p>AI agents choose an MCP tool by reading three things: its name, its description and its input schema. Many servers write those descriptions like docstrings, for a human who already knows why they opened the file. An agent with thirty tools in its context doesn't know that. It …
dev.to — MCP tag
TIER_1English(EN)·Rohan Sen Sharma·
<p>MCP vs API is rarely an either-or choice: MCP is a protocol that lets an AI agent discover and call tools at runtime, and an API is usually the service those tools wrap, so most teams building agents end up with both. The real decision is narrower. When does an agent need its …
<h1> Building Zero-Trust Model Context Protocol Servers: 10 Hardened MCP Tools with x402 Micropayments </h1> <p>The Model Context Protocol (MCP) by Anthropic is quickly becoming the standard interface for connecting frontier LLMs to tools, databases, and APIs. But as autonomous a…
<p><em>Pennyforge · dated evidence report, 2026-10-07 · follow-up to our <a href="https://dev.to/pennyforgehq/">dated MCP security timeline</a></em></p> <p>The Model Context Protocol has been around since November 2024, and it moved fast: the official TypeScript/Python SDK alone …
dev.to — MCP tag
TIER_1English(EN)·Royal Simpson Pinto·
<p>I have been building an MCP server called Bridgekit, and the part that took the longest was not exposing tools. Exposing tools is easy. The hard part was making sure a tool could only ever touch what a specific client was allowed to touch, and being able to prove afterward wha…
Medium — Claude tag
TIER_1English(EN)·Arjun Patel·
<p>I wanted a small, real MCP server to learn the protocol properly: not a "hello world" tool, but something an assistant could use for an actual task. French public data is a good playground. Two official APIs answer most "who is this company?" questions, need no key, and are pu…
<p>An MCP endpoint can return HTTP 200 while the integration still fails. A successful HTTP request does not establish that the response matches the JSON-RPC request, that initialization negotiated a supported revision, or that subsequent requests carry the session headers.</p> <…
<blockquote> <p><strong>TL;DR:</strong> In <a href="https://blog.bytechef.io/blogs/embedded-component-kit-chat" rel="noopener noreferrer">part three</a>, your own route fetched the user's tools from ByteChef and adapted each one to the AI SDK. The <strong>MCP Chat</strong> lets a…
dev.to — MCP tag
TIER_1English(EN)·Sospeter Mong'are·
<p>If you work in backend engineering, integrations, or AI development, you have probably started hearing about <strong>MCP</strong> alongside something you already know well: <strong>APIs</strong>.</p> <p>At first glance, MCP can look like another way of building an API. It is n…
<p>Chat assistants are good at conversation and bad at keeping a computer around. Close the tab and the environment is gone. Our product runs AI agents that each have their own persistent cloud machine: repos, tools, running apps, still there tomorrow.</p> <p>This week we put an …
<h2> Model access controls what an AI system can know and generate. Tool access controls what it can do. </h2> <p>The previous posts in The AI Control Plane have moved progressively closer to the point where policy becomes action. <a href="https://dev.to/kenwalger/ai-governance-f…
<p><strong>Following one question through the MCP stack — from the user's prompt to the final answer</strong></p> <p>Modern AI applications increasingly sit between two different worlds: <strong>MCP (Model Context Protocol)</strong> and an <strong>LLM provider API</strong>.</p> <…
Medium — MCP tag
TIER_1English(EN)·Ignacio Gonzalez Bohorquez·
<p>I use Upwork's official MCP server from Claude Code, and doceval, my open-source eval tool, ships an MCP server of its own. This is how MCP works underneath, and four things I learned along the way. Where I tested something, I say how.</p> <h2> MCP in one paragraph </h2> <p>MC…
Medium — MCP tag
TIER_1English(EN)·Ignacio Gonzalez Bohorquez·
<p>An MCP server's guards are the only code in the project where the test that matters asserts that nothing happened. No row came back. No request left the box. No file was opened. That is an awkward thing to assert, and it is the reason a guard suite can be green while proving a…
<p>Exposing one internal tool over the Model Context Protocol is a weekend project. Exposing ninety of them to humans <em>and</em> agents, without turning the MCP endpoint into a credential dump and the audit log into a fire hazard, is a platform problem.</p> <p>This post is the …
dev.to — MCP tag
TIER_1English(EN)·Shakar Bisetty·
<p><em>Part 4 of 10 · Building an Agentic Change-Approval MVP on MuleSoft</em></p> <p><a href="https://dev.to/thasha/part-3-designing-the-agent-network-and-what-the-brokers-graph-actually-looks-like-jap">Part 3</a> drew the broker's graph. Four of its nine nodes are MCP tool call…
dev.to — MCP tag
TIER_1English(EN)·ScriptMasterLabs·
<p>Safari 27 (released September 17, 2026) is the first major browser with a built-in Model Context Protocol server. Apple announced it on the WebKit blog (Saron Yitbarek) after debuting it in Safari Technology Preview 247 on July 1, 2026.</p> <h2> Setup </h2> <ol> <li>Safari >…
<p>A single MCP server in front of one service is a solved problem: generate tools from the OpenAPI spec, run it over stdio or HTTP, done. At company scale the problem changes shape. A midsize platform team has forty services, each with its own spec, its own auth, its own staging…
<p>Testing an MCP server through an AI client is a slow way to work. You type a natural-language prompt, hope the model picks the right tool, watch it fill in arguments, and if anything fails you cannot tell whether the bug is in your server, the model's reasoning, or the prompt.…
<p>A local MCP server launched over stdio inherits whatever credentials are already on your machine. Your shell has <code>AWS_PROFILE</code>, <code>DOCKER_HOST</code>, and a dozen tokens in <code>~/.config</code>, and the server just uses them. That is why most teams discover MCP…
<p>The API is done. Scenarios pass against staging, the mock kept three teams alive during development, and coding agents are calling the contract over MCP instead of guessing field names. Then a partner asks the question that quietly turns into a quarter-long project: "Can you s…
<p><em>Disclosure: I work on Pink Agentic AI Payments at PinkWallet; it's used as the example in the last section.</em></p> <p><strong>Short answer:</strong> every client below can send a custom header (or a key baked into the URL) to a remote Streamable HTTP MCP server — except …
dev.to — MCP tag
TIER_1English(EN)·Saptarshi Paul·
<p>Most software still has no API. Design tools, hardware control panels, the iOS Simulator, old enterprise apps, and half the settings screens on your laptop only speak mouse and keyboard. That is exactly where AI agents used to stop being useful. Computer use MCP servers close …
Medium — MCP tag
TIER_1English(EN)·Au Yong Jin Yoo·
<p>I started this project with a question: if a company already has APIs, how much work should it take to let an AI assistant use them?</p> <p>Say a company has a support-ticket API. People use it through an application to look up tickets and check their status. Now the company w…
<h2> Why I Built This </h2> <p>I use Claude, Cursor, and ChatGPT daily. I wanted them to control my Windows machine — run terminal commands, read files, search code, spawn background tasks.</p> <p>The problem? Most MCP servers for Windows were either:</p> <ul> <li> <strong>Cross-…
<div class="medium-feed-item"><p class="medium-feed-snippet">If you’ve been building with AI agents or LLMs recently, you’ve likely noticed a painful friction point:</p><p class="medium-feed-link"><a href="https://medium.com/@neerajgrg93/model-context-protocol-mcp-e…
Medium — MCP tag
TIER_1English(EN)·Shiva Praneeth Palla·
<p>A mid-size enterprise MCP deployment typically places a gateway in front of <strong>3 to 50 backend MCP servers</strong>, according to this <a href="https://aitutorialmaker.com/knowledge/what_does_an_enterprise_mcp_security_architecture_look_like_in_2026.php" rel="noopener nor…
Medium — MCP tag
TIER_1English(EN)·Tejas Kanazriya·
<!-- SC_OFF --><div class="md"><p>I’m to a point where I want to start opening up my platform and wondering what people are using for exposing APIs/MCPs for customers to use with their platforms.</p> <p>Is there anything out there that is reasonably cost for a startup that can gi…
