MCP Client Guides
The Prometheux Remote MCP Server works with any AI agent that supports the Model Context Protocol. Below you'll find step-by-step connection guides for specific clients.
Supported Clients
Coding Agents
We recommend using the Local MCP Server for coding agents — it uses stdio transport with direct credentials, which is more reliable than OAuth in IDE environments.
| Client | Description | Guide |
|---|---|---|
| Cursor | AI-powered code editor — connect via local MCP server | Connect Cursor → |
| Claude Code | Anthropic's terminal-based coding agent — connect via local or remote MCP server | Connect Claude Code → |
Chat & Enterprise Clients
These clients connect via the Remote MCP Server using OAuth 2.0 authentication.
| Client | Description | Guide |
|---|---|---|
| Claude | Anthropic's AI assistant — connect via Desktop or Web | Connect Claude → |
| ChatGPT | OpenAI's AI assistant — connect via Developer Mode connectors | Connect ChatGPT → |
| Genie Code | Databricks' autonomous AI agent for data work — connect via external MCP server registration | Connect Genie Code → |
| Snowflake Cortex | Snowflake's Cortex Code and Snowflake Intelligence — connect via external MCP connector | Connect Snowflake Cortex → |
Other Clients
Any MCP-compatible client can connect to Prometheux. For stdio-capable clients (IDEs, terminal agents), we recommend the Local MCP Server. For browser-based clients, use the Remote MCP Server with OAuth 2.0.
Prerequisites
All client connections require:
- Prometheux account with valid credentials
- Username, organization, and API token from your Prometheux admin
- For coding agents: the
prometheux-mcppackage (installation instructions) - For browser-based clients: the remote MCP server at
https://api.prometheux.ai/px-remote-mcp-server