> ## Documentation Index
> Fetch the complete documentation index at: https://mcp-use.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.
## Submitting Feedback
If you encounter incorrect, outdated, or confusing documentation on this page, submit feedback:
POST https://mcp-use.com/docs/feedback
```json
{
"path": "/inspector/index",
"feedback": "Description of the issue"
}
```
Only submit feedback when you have something specific and actionable to report.
# Inspector
> Built-in web interface for real-time server monitoring and debugging
> For the complete documentation index, see [llms.txt](https://manufact.com/docs/llms.txt).
[
](https://inspector.mcp-use.com)
The MCP Inspector is a powerful web-based debugging and inspection tool for
MCP (Model Context Protocol) servers. It provides a beautiful, intuitive
interface for testing tools, exploring resources, managing prompts, and
monitoring server connections - all from your browser.
### Key Features
* **Multi-Server Management**: Connect to and manage multiple MCP servers simultaneously
* **Tool Testing**: Interactive tool execution with schema validation and real-time results
* **Resource Browser**: Explore available resources and data sources
* **Prompt Manager**: Test and manage prompts with argument templates
* **Chat Interface**: Interactive chat with LLM integration for testing conversational flows
* **OAuth Support**: Built-in OAuth flow handling with popup authentication
* **Session Persistence**: Connections saved to localStorage with auto-reconnect
* **OpenAI Apps SDK Support**: Full widget rendering and debugging for ChatGPT Apps
* **Saved Requests**: Save and replay tool executions
* **Command Palette**: Powerful search and navigation with `Cmd/Ctrl + K`
* **Add to Client**: Quickly add MCP servers to Cursor, VS Code, Claude Desktop, and CLI tools
## Documentation
Explore the inspector documentation:
### Getting Started
* **[CLI Usage](/inspector/cli)** - Run the inspector from the command line
* **[Connection Settings](/inspector/connection-settings)** - Advanced configuration options
* **[URL Parameters](/inspector/url-parameters)** - Deep links and auto-connect
### Features
* **[Overview](/inspector/index)** - Saved requests, preview mode, and connection management
* **[Keyboard Shortcuts](/inspector/keyboard-shortcuts)** - Complete shortcuts reference
* **[Command Palette](/inspector/command-palette)** - Search and navigation guide
* **[RPC Logging](/inspector/rpc-logging)** - Monitor JSON-RPC messages in real-time
### Integration & Advanced
* **[Integration](/inspector/integration)** - Mount inspector on Express or Hono apps
* **[Debugging ChatGPT Apps](/inspector/debugging-chatgpt-apps)** - Test and debug ChatGPT Apps with OpenAI Apps SDK
* **[Self-Hosting](/inspector/self-hosting)** - Deploy your own inspector instance
* **[Development](/inspector/development)** - Development setup and contribution guide
* **[Changelog](/inspector/changelog)** - Version history and updates
# Getting Started
## Dashboard
The MCP Inspector dashboard is your central hub for managing MCP server connections.
What you see:
* Connected Servers panel (left)
* Connect panel with transport, URL, connection type, auth, headers (right)
* Quick actions in the top bar
Next: add the Linear MCP server to begin a live walkthrough.
### Add MCP Server
Open the Connect panel on the right side of the dashboard.
Set Transport Type to "Streamable HTTP" for SSE connections.
Enter the MCP server URL (e.g., `https://mcp.linear.app/mcp`).
Click the Connect button to establish the connection.
### OAuth Authentication
All authentication tokens and credentials are stored securely in your
browser's local storage. Nothing is sent to our servers - everything stays on
your device.
Servers like Linear require OAuth. After clicking Connect, you'll see the authorization page.
Click "Approve" to grant access. The inspector handles the redirect automatically.
### Connected Server
Once authenticated, the server appears in the Connected Servers list with a green indicator.
Click the icon buttons to Inspect, Disconnect, or Remove the server.
# Server Detail View
Clicking "Inspect" opens the detailed server view with tabs for Tools, Resources, Prompts, and Chat.
### Tools Tab
The Tools tab displays all available tools from the MCP server. Each tool shows its name and description.
Select a tool to view its details and execute it. The right panel will show the tool's schema and an Execute button.
### Resources Tab
The Resources tab displays available resources from the MCP server. Resources are data sources the server can provide.
Note: Linear MCP server doesn't expose any resources, so this tab shows "No resources available".
### Prompts Tab
The Prompts tab shows pre-configured prompts from the MCP server. You can select and execute prompts with parameters.
Note: Linear MCP server doesn't provide any prompts, so this tab shows "No prompts available".
### Elicitation Tab
The Elicitation tab appears when a tool requests user input during execution.
What you can do:
* Review pending elicitation requests
* Fill form fields generated from the server's schema
* Accept, decline, or cancel requests
* Jump back to tool results after response submission
Supported schema patterns include all SEP-1330 enum variants:
* Untitled single-select: `type: "string"` + `enum`
* Titled single-select: `type: "string"` + `oneOf[{ const, title }]`
* Legacy titled enum: `type: "string"` + `enum` + `enumNames`
* Untitled multi-select: `type: "array"` + `items.enum`
* Titled multi-select: `type: "array"` + `items.anyOf[{ const, title }]`
These schemas render as dropdowns (single-select) or checkbox groups (multi-select), while submitted values always use the underlying enum/const values.
# Chat Tab
The Chat feature uses a BYOK (Bring Your Own Key) model. Your API key is
stored locally in your browser and never sent to our servers. All requests are
made directly from your device to your LLM provider.
The Chat tab provides an interactive interface to test the MCP server with an LLM. Configure your API key to start chatting.
### Chat Setup
To use the chat:
Click "Configure API Key" to open the configuration modal.
* **Provider**: Select your LLM provider (OpenAI, Anthropic, etc.)
* **Model**: Choose the model to use (gpt-4o, claude-3, etc.)
* **API Key**: Enter your API key (stored locally in browser)
* Click "Save Configuration" to activate the chat
Once configured, the chat interface becomes active.
Type natural language queries to test the MCP server.
See tool calls and responses in real-time with detailed JSON output.
When you send a message, the inspector shows real-time tool calls and responses.
Key features of the chat interface:
* **Message Input**: Natural language queries to test MCP server functionality
* **Tool Call Visualization**: Shows which tools are being called with indicators
* **Tool Call Details**: Side panel displays arguments and results in JSON format
* **Response Display**: Shows the assistant's response based on tool results
* **Real-time Feedback**: See exactly how the MCP server processes requests
### Using Prompts in Chat
If your MCP server provides prompts, you can invoke them directly from the chat interface. This allows you to execute prompts and include their results in your message context before sending.
Prompts with arguments are not supported yet. Only prompts without parameters
can be executed from the chat interface.
Type `/` in the chat input field to open the prompts dropdown menu.
The dropdown displays all available prompts from the connected MCP server.
Navigate through available prompts using:
* **Arrow Keys** (↑/↓): Move between prompts
* **Mouse Click**: Select a prompt directly
* **Enter**: Execute the focused prompt
* **Escape**: Close the dropdown without selecting
Each prompt shows its name and description to help you choose the right one.
When you select a prompt, it executes immediately and its result is included in your message context.
The prompt result appears in the chat input area, and you can continue typing your message or send it directly.
Key points about prompts in chat:
* **Arguments Not Supported**: Only prompts without parameters are currently supported
* **Results Included First**: Prompt results are added to your message context before sending
* **Keyboard Navigation**: Use arrow keys, Enter, and Escape for quick selection
* **Inline Execution**: Prompts execute immediately upon selection without blocking the chat
# Add to Client
The inspector includes a "Add to Client" button in the header that allows you to quickly add the current MCP server to various AI clients and code editors.
## Keyboard Shortcuts
The inspector includes powerful keyboard shortcuts for quick navigation:
**Cmd/Ctrl + K** opens the quick search modal where you can:
* Quickly connect to a new server
* Access documentation and tutorials
* Jump to recently used servers
* Search and execute tools directly
* Navigate to different tabs (Tools, Prompts, Resources, Chat)
For a complete list of all keyboard shortcuts, see the [Keyboard Shortcuts](/inspector/keyboard-shortcuts) guide.
## Persistency
The inspector automatically saves your server configurations to localStorage:
* Server connections persist across page reloads
* Credentials are stored securely in your browser
* Clear all sessions with the "Clear All" button
Visit [inspector.mcp-use.com](https://inspector.mcp-use.com) to start inspecting your MCP servers today!