Other MCP clients
The Zuplo MCP Gateway works with any MCP client that speaks the Streamable HTTP transport. The pages in this section give the basics for several popular clients beyond the ones with dedicated guides (Claude Desktop, Claude Code, ChatGPT, Cursor, VS Code).
For each client, the pattern is the same: paste your MCP route URL
(https://{deploymentUrl}/{routePath}, for example
https://{deploymentUrl}/mcp/linear-v1) into the client's configuration, then
complete the gateway's OAuth flow in your browser when prompted.
Windsurf
Windsurf is an agentic IDE from Codeium with a built-in AI flow called Cascade. Cascade supports stdio, Streamable HTTP, and the older SSE transport, with OAuth available on each.
To add the gateway, edit ~/.codeium/windsurf/mcp_config.json and add an entry
under mcpServers:
~/.codeium/windsurf/mcp_config.json
Windsurf supports environment-variable interpolation with ${env:VAR_NAME} in
the config so you don't have to hardcode the URL or any header values.
Restart Cascade and complete the gateway's OAuth flow in your browser when prompted.
Capabilities supported: tools.
Docs: Windsurf Cascade MCP.
Goose
Goose is Block's open-source AI agent. Goose supports the Streamable HTTP transport for remote MCP servers — it calls them remote extensions.
To add the gateway from the Goose CLI:
Code
Choose Add Extension → Remote Extension (Streamable HTTP) and follow the prompts. Enter your route URL when asked. From the Goose Desktop app, open Settings → Extensions → Add custom extension and paste the same URL.
Goose then opens the gateway's OAuth flow in your browser. Sign in and complete the upstream consent page.
Capabilities supported: tools, prompts, resources, roots, sampling, elicitation, MCP Apps.
Docs: Goose extensions.
Gemini CLI
The Gemini CLI is Google's
open-source terminal agent for Gemini. Configure MCP servers in
~/.gemini/settings.json under the mcpServers key:
~/.gemini/settings.json
Restart the CLI. The Gemini CLI registers itself with the gateway through Dynamic Client Registration and opens the OAuth flow in your browser on first connect.
Capabilities supported: tools, prompts.
Docs: Gemini CLI MCP server configuration.
GitHub Copilot CLI
The GitHub Copilot CLI is GitHub's
terminal agent. Use the /mcp add slash command from inside a Copilot CLI
session to add the gateway. Copilot CLI prompts for the server name, transport
(http), and URL, then writes the entry to ~/.copilot/mcp-config.json and
starts the OAuth flow.
Capabilities supported: tools, sampling, elicitation, OAuth Client Credentials.
Docs: Add an MCP server.
Postman
Postman supports MCP server testing and debugging through its AI Agent Builder. Create a new MCP request, choose Streamable HTTP, and paste your MCP route URL. Postman handles the OAuth flow inline.
Postman is a useful client for verifying that an MCP route responds correctly, inspecting raw JSON-RPC traffic, and debugging tool schemas.
Capabilities supported: tools, prompts, resources, sampling, elicitation, MCP Apps.
Docs: Postman download (MCP support is built into the standard Postman desktop client).
MCPJam
MCPJam Inspector is an open-source local development client for MCP servers, ChatGPT apps, and MCP App extensions. It's a strong tool for verifying gateway behavior end-to-end during development.
Launch the inspector, set the transport to Streamable HTTP, and paste your MCP route URL. MCPJam runs the OAuth flow and shows tools, prompts, resources, and any MCP Apps the gateway exposes.
Capabilities supported: tools, prompts, resources, elicitation, MCP Apps, DCR, CIMD, Tasks.
Docs: MCPJam getting started.
Continue
Continue is an open-source AI code assistant that runs in VS Code and JetBrains IDEs. Continue supports MCP servers in agent mode only.
Add the gateway either by editing your main Continue config or by creating a
standalone YAML file under .continue/mcpServers/. For a remote HTTP server:
.continue/mcpServers/zuplo.yaml
Capabilities supported: tools, prompts, resources, MCP Apps.
Docs: Continue MCP deep dive.
LibreChat
LibreChat is an open-source self-hosted AI chat UI. LibreChat supports MCP servers as additional tool sources.
Add the gateway in the mcpServers section of your librechat.yaml
configuration, restart the LibreChat server, and complete the OAuth flow the
first time you select the connector in a conversation.
Capabilities supported: tools.
Docs: LibreChat MCP.
JetBrains AI Assistant
The JetBrains AI Assistant plugin is available in every JetBrains IDE. It supports MCP servers as sources of tools for the AI Assistant.
Open the AI Assistant settings, add an MCP server, and paste your MCP route URL. Complete the OAuth flow in your browser when prompted.
Capabilities supported: tools.
Docs: Model Context Protocol in JetBrains AI Assistant.
Other clients
For an up-to-date list of MCP clients, see the official MCP clients page. Any client that supports the Streamable HTTP transport and standard MCP authorization (RFC 9728 Protected Resource Metadata plus OAuth 2.1) will work with the Zuplo MCP Gateway out of the box.