For the MCP ecosystem to grow, we need better developer tools. Today, we're publicly launching MCP Inspector.
The Model Context Protocol (MCP) is rapidly becoming the standard for connecting AI assistants to external tools and data sources. But as developers build MCP servers, they face a common challenge: how do you test and debug them effectively?
Until now, testing meant setting up local environments, managing dependencies, or logging into platforms that collect your data. We built MCP Inspector to change that.
What is MCP Inspector?
MCP Inspector is a free, browser-based tool that lets you connect to any MCP server URL and interact with its full capabilities—tools, resources, prompts, and tasks—directly from your browser.
Try it now: glama.ai/mcp/inspector
Why We Built This
When we started building MCP integrations, we found ourselves constantly switching between terminals, debugging configurations, and writing throwaway scripts just to test a single tool call. The official MCP Inspector is great for local development, but we needed something we could use anywhere—to test remote servers, share debugging sessions with teammates, or quickly verify a deployment.
So we built what we needed: a zero-friction inspector that runs entirely in your browser.
Key Features
No Login Required
Open the URL, paste your server address, and start inspecting. No account creation, no signup flow, no barriers.
Privacy-First Design
All requests go directly from your browser to the MCP server. We don't proxy, log, or store any of your requests or responses. Your API keys and server interactions stay between you and your server.
Full Protocol Support
We didn't cut corners. MCP Inspector supports the complete MCP specification:
- Tools — List available tools, configure parameters with a dynamic form, and execute them
- Resources & Templates — Browse and read server resources
- Prompts — Test prompt templates with arguments
- Tasks — Create, monitor, cancel, and retrieve results from long-running tasks
- Progress Notifications — See real-time progress updates during execution
- Elicitations — Respond to server-initiated form requests
- OAuth 2.0 — Full OAuth flow support with dynamic client registration
- Bearer Tokens & Custom Headers — Flexible authentication options
Built-in Demo Server
Not sure how it works? We've included a test server (mcp-test.glama.ai/mcp) that demonstrates every feature—tasks, elicitations, progress notifications, audio responses, and images. It's the fastest way to understand what MCP can do.
Shareable Sessions
Your entire configuration—servers, selected tools, and arguments—is stored in the URL. Bookmark it to save your setup, or share the link with a colleague to give them the exact same view.
How to Use It
- Go to glama.ai/mcp/inspector
- Click "Add Server" and enter your MCP server URL
- Select your authentication method (None, OAuth, Bearer Token, or Custom Headers)
- Connect and start exploring
The interface shows your server's capabilities in tabs: Tools, Resources, Resource Templates, Prompts, and Tasks. Select any item to see its details, configure parameters, and execute requests. All requests and responses are logged in the panel below, and additional debugging info is available in your browser's developer console.
For Local Development
MCP Inspector works great for testing remote servers. For local development with stdio transports or other local-only features, we recommend the official MCP Inspector which is designed specifically for that workflow.
Try It Today
We've been using MCP Inspector internally for months, and it's become an essential part of our development workflow. We're excited to share it with the community.
Have feedback or feature requests? Join our Discord and let us know what you'd like to see.
The MCP Inspector is part of Glama's suite of tools for the Model Context Protocol ecosystem. Explore more at glama.ai/mcp.
Top comments (0)