Remote MCP Server
The remote MCP server (px-remote-mcp-server) is a cloud-hosted Model Context Protocol server that enables any MCP-compatible AI agent to interact with Prometheux — without needing to install anything locally.
It uses OAuth 2.0 authentication with PKCE, so users log in through a browser and their credentials are never exposed to the AI client.
Why Use Remote MCP?
Best for: Multi-client integrations, enterprise environments, browser-based AI agents
Pros:
- Works with any MCP-compatible client (Claude, Genie Code, and more)
- OAuth 2.0 authentication with browser login flow
- Centralized deployment (one server, many users)
- Handles authentication and session management
- No local installation required
Cons:
- Requires a deployed server instance
- Suitable for production cloud deployments
If you prefer a local installation that doesn't require a deployed server, check out the Local MCP Server documentation.
Server Details
The Prometheux remote MCP server is deployed at:
| Property | Value |
|---|---|
| Base URL | https://api.prometheux.ai/px-remote-mcp-server |
| Host | https://api.prometheux.ai |
| Port | 443 |
| Base path | /px-remote-mcp-server |
| Protocol | Streamable HTTP (MCP over HTTP + SSE) |
| Authentication | OAuth 2.0 Authorization Code + PKCE |
OAuth Endpoints
| Endpoint | URL |
|---|---|
| Authorization | https://api.prometheux.ai/px-remote-mcp-server/oauth/authorize |
| Token | https://api.prometheux.ai/px-remote-mcp-server/oauth/token |
| User Info | https://api.prometheux.ai/px-remote-mcp-server/oauth/userinfo |
| Discovery | https://api.prometheux.ai/px-remote-mcp-server/.well-known/oauth-authorization-server |
OAuth Parameters
When configuring a client manually, use these parameters:
| Parameter | Value |
|---|---|
| Client ID | prometheux |
| OAuth Scope | openid profile |
| Authorization Endpoint | https://api.prometheux.ai/px-remote-mcp-server/oauth/authorize |
| Token Endpoint | https://api.prometheux.ai/px-remote-mcp-server/oauth/token |
Authentication Flow
When connecting through any MCP client, the OAuth flow works as follows:
- The client redirects you to the Prometheux login page
- You enter your credentials:
- Username: Your Prometheux username
- Organization: Your Prometheux organization
- API Token: Your Prometheux authentication token
- After successful authentication, the client receives a session token
- All subsequent MCP requests use this token automatically
Connect Your Client
We provide step-by-step guides for connecting specific MCP clients:
| Client | Guide |
|---|---|
| Claude (Desktop & Web) | Connect Claude → |
| ChatGPT (OpenAI) | Connect ChatGPT → |
| Genie Code (Databricks) | Connect Genie Code → |
| Snowflake Cortex (Cortex Code & Intelligence) | Connect Snowflake Cortex → |
| Cursor | Connect Cursor → |
| Claude Code (Terminal CLI) | Connect Claude Code → |
For coding agents like Cursor and Claude Code, we recommend using the Local MCP Server instead. It uses stdio transport with direct credentials, which avoids the OAuth browser flow and is more reliable in IDE and terminal environments.
For other MCP-compatible clients, use the Server Details and OAuth Endpoints above to configure the connection manually.
Session Management
- Sessions persist until you explicitly disconnect or the server restarts
- Re-authentication: If your session expires, the client will prompt you to log in again
- Security: Your Prometheux token remains valid until you revoke it from your Prometheux dashboard
Troubleshooting
"Unable to connect" error
The client can't reach the remote MCP server.
Solution:
- Verify the server URL is correct:
https://api.prometheux.ai/px-remote-mcp-server - Check if your Prometheux credentials are valid
- Ensure the remote MCP server is running and accessible
- Try refreshing and reconnecting
OAuth login page doesn't load
The authentication flow isn't starting properly.
Solution:
- Check your browser's popup blocker settings
- Try the connection flow in an incognito/private window
- Verify the remote MCP server is running and accessible
- Clear your browser cache and cookies
"Authentication required" after login
Your session expired or credentials are invalid.
Solution:
- Your Prometheux token may have been revoked or expired
- Re-authenticate by disconnecting and reconnecting
- Contact your Prometheux admin to verify your credentials
- Check that you entered the correct username, organization, and token
Tools not appearing
The AI agent doesn't recognize the Prometheux tools.
Solution:
- Ensure you completed the OAuth flow successfully
- Try disconnecting and reconnecting the server
- Verify the connection shows as active in your client's settings
- Be more explicit in your query: "Use the prometheux tools to list concepts in project X"
"Connection timeout" error
The remote MCP server is taking too long to respond.
Solution:
- Check your internet connection
- Verify the Prometheux instance is accessible and running
- Check if you're behind a VPN or firewall that might be blocking access
- Try again in a few moments — the server might be temporarily overloaded
Learn More
- MCP Protocol Specification: modelcontextprotocol.io
- Available Tools: See MCP Overview for the full tool reference
- Local MCP Server: See Local MCP Server for Claude Desktop integration
- Python SDK: See Python SDK Reference for alternative integration methods
- REST API: See REST API Reference for direct HTTP access
Related Resources
- Concepts API - Learn about the concept API
- Python SDK - Alternative programmatic access
- Chat API - Interactive AI chat interface