--- title: Agents Can Curl, So Why Do I Need an MCP Server? | daily.dev description: An argument for when a Model Context Protocol server is worth building versus when it is unnecessary overhead. If an API is clean, well-described via OpenAPI,... canonical: https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Agents Can Curl, So Why Do I Need an MCP Server? | daily.dev og:description: An argument for when a Model Context Protocol server is worth building versus when it is unnecessary overhead. If an API is clean, well-described via OpenAPI,... og:url: https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8 og:image: https://api.daily.dev/og/posts/LQrVN7xA8.png og:image:alt: Agents Can Curl, So Why Do I Need an MCP Server? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Agents Can Curl, So Why Do I Need an MCP Server? **[API Evangelist](https://daily.dev/sources/apievangelist)** · 5 min read · 0 upvotes · 0 comments ## Summary An argument for when a Model Context Protocol server is worth building versus when it is unnecessary overhead. If an API is clean, well-described via OpenAPI, and the job is a single stateless call, an agent can curl it directly with no server needed. MCP servers earn their place in four scenarios: multi-step orchestration, narrowing/framing a large surface for a specific task, wrapping non-REST or legacy systems agents cannot curl, and adding guardrails/control between an improvising model and production systems. The core claim is that MCP servers never create capabilities themselves — they only add orchestration, framing, reach, or control on top of a real API — and if you cannot tell whether you need one, the real problem is that your API isn't clean enough yet. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Do I need to build an MCP server if my AI agent can already call my API directly via curl and OpenAPI? Not necessarily. If the API is clean, has a good OpenAPI definition, sensible auth, and the job is a single stateless call, an agent can curl it directly and an MCP server would just be unnecessary overhead — a second contract that can drift from the original API. An MCP server earns its place only when the job needs multi-step orchestration, a narrowed and framed subset of tools, access to non-REST or legacy systems, or guardrails and access control between the agent and production. _Teams weighing MCP servers against direct API calls can compare real-world agent architecture takes on daily.dev._ ### What are valid reasons to build an MCP server for an AI agent instead of letting it call the API directly? Four cases justify it: state and multi-step work, where an MCP server presents a five-call choreography as one named tool instead of leaving the agent to reconstruct it; affordance and framing, narrowing two hundred flat OpenAPI operations down to the six relevant ones for a task; reaching non-REST or legacy systems an agent could never curl on its own; and control, where the server acts as a guardrail and audit layer between an improvising model and a production system. _Engineers designing agent-facing infrastructure can track practical MCP architecture guidance on daily.dev._ ## Similar posts on daily.dev - [Most MCP servers don't need to exist. Your case might be an exception.—Martian Chronicles, Evil Martians’ team blog](https://daily.dev/posts/most-mcp-servers-don-t-need-to-exist-your-case-might-be-an-exception-martian-chronicles-evil-mart-y1jdqnyqv) · Martian Chronicles · 1 upvotes · 0 comments - [MCP Servers: Your Next User Is An AI Agent](https://daily.dev/posts/mcp-servers-your-next-user-is-an-ai-agent-apyzw7sof) · marmelab · 2 upvotes · 0 comments - [Why the MCP Server Is Now a Critical Microservice](https://daily.dev/posts/why-the-mcp-server-is-now-a-critical-microservice-gi57gs1tb) · The New Stack · 1 upvotes · 0 comments - [Building agents that reach production systems with MCP](https://daily.dev/posts/building-agents-that-reach-production-systems-with-mcp-zz2sr0sr5) · Claude · 3 upvotes · 0 comments --- Tags: [#architecture](https://daily.dev/tags/architecture), [#ai-agents](https://daily.dev/tags/ai-agents), [#mcp](https://daily.dev/tags/mcp), [#openapi](https://daily.dev/tags/openapi) [View this post on daily.dev](https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Agents Can Curl, So Why Do I Need an MCP Server?","url":"https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8"},"datePublished":"2026-08-20T11:55:01.043Z","dateModified":"2026-08-20T11:55:51.402Z","description":"An argument for when a Model Context Protocol server is worth building versus when it is unnecessary overhead. If an API is clean, well-described via OpenAPI,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ace92ba45185da574423a424f0ba6a73?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ace92ba45185da574423a424f0ba6a73?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"API Evangelist","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"API Evangelist","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/a36244ae67bc4f41a605f780267ecb5f","url":"https://daily.dev/sources/apievangelist"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"architecture,ai-agents,mcp,openapi","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"API Evangelist","item":"https://daily.dev/sources/apievangelist"},{"@type":"ListItem","position":3,"name":"Agents Can Curl, So Why Do I Need an MCP Server?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/agents-can-curl-so-why-do-i-need-an-mcp-server--lqrvn7xa8#faq","mainEntity":[{"@type":"Question","name":"Do I need to build an MCP server if my AI agent can already call my API directly via curl and OpenAPI?","acceptedAnswer":{"@type":"Answer","text":"Not necessarily. If the API is clean, has a good OpenAPI definition, sensible auth, and the job is a single stateless call, an agent can curl it directly and an MCP server would just be unnecessary overhead — a second contract that can drift from the original API. An MCP server earns its place only when the job needs multi-step orchestration, a narrowed and framed subset of tools, access to non-REST or legacy systems, or guardrails and access control between the agent and production. Teams weighing MCP servers against direct API calls can compare real-world agent architecture takes on daily.dev."}},{"@type":"Question","name":"What are valid reasons to build an MCP server for an AI agent instead of letting it call the API directly?","acceptedAnswer":{"@type":"Answer","text":"Four cases justify it: state and multi-step work, where an MCP server presents a five-call choreography as one named tool instead of leaving the agent to reconstruct it; affordance and framing, narrowing two hundred flat OpenAPI operations down to the six relevant ones for a task; reaching non-REST or legacy systems an agent could never curl on its own; and control, where the server acts as a guardrail and audit layer between an improvising model and a production system. Engineers designing agent-facing infrastructure can track practical MCP architecture guidance on daily.dev."}}]} ```