Postman launches full support for Model Context Protocol (MCP) — Build better AI Agents, faster

Today, we’re thrilled to announce integrated support for the Model Context Protocol (MCP) directly in Postman. This marks a big step forward in our investment in AI Agent Builder — our solution for making it easier to design, debug, and deploy AI agents using the tools and APIs you already know and trust.

With this release, you can now create and send MCP requests using the familiar Postman API client and generate MCP servers from our network of 100,000 APIs. Whether you’re calling a tool, composing a prompt, exploring a server’s available resources, or looking to build your server, it all works seamlessly inside Postman — no setup, no scripts, no guesswork. Just faster, more reliable agent development.

A familiar interface for a powerful new protocol

The Model Context Protocol (MCP) streamlines API integrations for LLMs and agentic frameworks by defining a consistent structure for exposing and invoking tools, prompts, and resources. It acts as a thin but powerful layer over APIs, standardizing how agents interact with underlying services across transports like STDIO, SSE, and streamable HTTP. But working with MCP directly can still be tricky—especially when it comes to debugging requests, validating configurations, and testing responses.

That’s where Postman comes in. Starting today, you can open a workspace, click “New,” and select “MCP” just like you would for any other type of request (HTTP, gRPC, GraphQL). From there, you can connect to any MCP server, explore its capabilities, and build and debug visually — no code or manual configuration required.

Postman handles the protocol-level details so you can focus on what matters: what your agent does and whether it works! Tools, prompts, and resources are auto-loaded from the server and displayed in a structured interface. You can even paste an MCP config JSON (from tools like Claude or VS Code) into the URL bar to generate a working request instantly.

Every MCP request can be saved, reused, shared, and documented — just like any other request type in Postman. That means you can collaborate with your team, test new ideas, or include MCP directly in your automated workflows with minimal friction.

Public API to MCP server in seconds — with AI Tool Builder

We’ve also launched the fastest way to go from APIs to fully functioning MCP tools: the AI Tool Builder in Postman.

Now you can generate MCP servers that include only the tools your application actually needs — no extra setup, no unnecessary code. These are built on Postman’s network of over 100,000 public APIs—including verified sources like Salesforce, UPS, and X—giving you a reliable foundation for dependable AI agents.

Just select the APIs you want, click Build, and download your custom server. Test it right away in the Postman MCP Client and use it in products like Claude, Cursor, and others.

 

Available today — and built for what’s next

All of these MCP improvements are integrated into Postman’s existing workflows and available starting today. Download the latest version of the Postman app, create a new MCP request, or generate an MCP server with AI Tool Builder and you’re off and building.

We’ll be showcasing our MCP solutions in hands-on demos and workshops at POST/CON 25. Curious about what’s next in AI agent development or want to learn how to build? We’d love to see you there. Register now!

And as always, your feedback helps shape the product. Join our #mcp-lab channel on Discord to share what you’re building and stay in the loop on our latest AI updates.