Netrefer.Lumen

Your affiliate platform, illuminated.

Uses your existing NetRefer account

Connect your AI tool

Claude Desktop

  1. Sign in above to generate a Bearer token for your account.
  2. Open your Claude Desktop config file:
    macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
    Windows: %APPDATA%\Claude\claude_desktop_config.json
  3. Add the following block inside "mcpServers":
{
  "mcpServers": {
    "netrefer-data": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://your-server/sse"]
    }
  }
}
No token required β€” mcp-remote opens a browser to sign in via your NetRefer account and caches the token automatically. Requires Node.js installed.

Claude Code (CLI)

  1. Sign in above to get your Bearer token.
  2. Run this command in your terminal to register the server:
claude mcp add --transport sse netrefer-data \
  https://your-server/sse \
  --header "Authorization: Bearer YOUR_TOKEN"

Or add it manually to ~/.claude.json (global) or .claude/settings.json (project):

{
  "mcpServers": {
    "netrefer-data": {
      "type": "sse",
      "url": "https://your-server/sse",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}
Claude Code CLI v1.0+ supports SSE MCP servers natively. The server is available in all Claude Code sessions once added.

Cursor

  1. Open Cursor Settings β†’ MCP (or edit ~/.cursor/mcp.json).
  2. Add a new MCP server entry:
{
  "netrefer-data": {
    "url": "https://your-server/sse",
    "headers": {
      "Authorization": "Bearer YOUR_TOKEN"
    }
  }
}
Cursor supports SSE transport. No local process needed β€” just the URL and token.

Windsurf (Codeium)

  1. Open Windsurf Settings β†’ Cascade β†’ MCP Configuration, or edit ~/.codeium/windsurf/mcp_config.json.
  2. Add the following server entry:
{
  "mcpServers": {
    "netrefer-data": {
      "serverUrl": "https://your-server/sse",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}
Windsurf uses serverUrl (not url) for remote MCP servers. Restart Cascade after saving the config.

Cline (VS Code Extension)

  1. In VS Code, open the Cline panel and click MCP Servers β†’ Remote MCP Server.
  2. Enter the server URL and token when prompted β€” or edit the config file directly:
    macOS/Linux: ~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
    Windows: %APPDATA%\Code\User\globalStorage\saoudrizwan.claude-dev\settings\cline_mcp_settings.json
{
  "mcpServers": {
    "netrefer-data": {
      "url": "https://your-server/sse",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}
Cline auto-discovers tools on connection. The autoApprove list controls which tools run without confirmation.

VS Code β€” GitHub Copilot

  1. Requires VS Code 1.99+ with the GitHub Copilot Chat extension.
  2. Create or edit .vscode/mcp.json in your workspace (or add to user settings.json):
{
  "servers": {
    "netrefer-data": {
      "type": "sse",
      "url": "https://your-server/sse",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}
  1. Open Copilot Chat (Ctrl+Alt+I), switch to Agent mode, and the netrefer-data tools will appear automatically.
Use .vscode/mcp.json for project-scoped config. Add it to .gitignore to avoid committing your token.

Continue (VS Code / JetBrains)

  1. Edit ~/.continue/config.yaml:
mcpServers:
  - name: netrefer-data
    transport:
      type: sse
      url: https://your-server/sse
      headers:
        Authorization: "Bearer YOUR_TOKEN"
  1. Reload the Continue extension β€” the tools appear in the chat context menu.
Continue supports both VS Code and JetBrains IDEs. Use the same config file for both.

n8n

  1. In your workflow, add an MCP Client Tool node (n8n v1.68+).
  2. Configure the node:
    • Transport: SSE
    • URL: https://your-server/sse
    • Authentication: Header Auth β†’ Authorization: Bearer YOUR_TOKEN
  3. Connect the MCP Client Tool to an AI Agent node β€” the query_data, list_datasets, and describe_dataset tools will be available automatically.

Alternatively, use an HTTP Request node to call the REST API directly:

URL:    https://your-server/api/chat
Method: POST
Header: Authorization: Bearer YOUR_TOKEN
Body:   { "question": "{{ $json.question }}" }
The REST /api/chat endpoint returns results as plain text β€” easier to wire into downstream nodes without MCP tooling.

OpenAI / ChatGPT Custom GPT

  1. In ChatGPT β†’ Explore GPTs β†’ Create, go to the Configure tab.
  2. Under Actions, click Add action and set the schema URL:
https://your-server/openapi.json
  1. Set Authentication to API Key, type Bearer, and paste your token.
  2. For the OpenAI Responses API or Assistants with tool use, point the MCP endpoint at:
https://your-server/sse
Pass your Bearer token in the Authorization header.

Custom Agent / API

Connect any MCP-compatible agent using SSE transport.

https://your-server/sse
GET /sse  HTTP/1.1
Host: your-server
Authorization: Bearer YOUR_TOKEN
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "query_data",
    "arguments": { "question": "Top affiliates by revenue last month" }
  },
  "id": 1
}
  • list_datasets β€” discover available tables
  • describe_dataset β€” get column schema for a table
  • query_data β€” natural language β†’ SQL β†’ results
Tenant isolation is automatic β€” results are always filtered to your account's subscriber GUIDs.

Available MCP Tools

list_datasets read-only idempotent

List all available datasets in the data catalog. Returns table names, column counts, and a preview of column names.

Parameters none
Returns List of dataset names with column counts
describe_dataset read-only idempotent

Get the full column schema for a specific dataset β€” column names, data types, and descriptions. Use this before querying to understand table structure.

Parameters
table_namestringExact table name as returned by list_datasets (e.g. ds_fact_clicks)
Returns Column names, types, and descriptions
list_workflows read-only idempotent

List active automation workflows available to your account, grouped by project. Includes each workflow's trigger endpoint and expected payload.

Parameters none
Returns Workflow names, IDs, endpoints, and payload descriptions
query_data read-only

Query affiliate data using a plain English question. Automatically selects relevant tables, generates SQL, executes it against Databricks, and returns results as a markdown table.

Parameters
questionstringPlain English question, e.g. "Top 10 affiliates by revenue this month"
tablesstring[] Β· optionalLimit query to specific tables. Auto-detected if omitted.
Returns Markdown table of results (up to 500 rows)
run_workflow

Execute an automation workflow by ID. Use list_workflows first to find the workflow ID and understand its expected input payload.

Parameters
workflow_idstringWorkflow ID from list_workflows
payloadstring Β· optionalJSON input for the workflow. Defaults to {}
Returns Workflow execution result as JSON
Tenant isolation is enforced automatically β€” all results are filtered to your account's subscriber GUIDs. You cannot query data outside your scope.
Chat
πŸ”‘ My Token
πŸ‘‹ Hi! Ask me anything about your affiliate data. Examples: β€’ "Show top 10 affiliates by revenue this month" β€’ "How many conversions happened last week?" β€’ "Compare campaign performance across subscribers"
Assistant

Table Visibility Configuration

Select which tables are exposed to users. Click βš™ on any table to configure its columns and descriptions.

Loading…

Browse Gold Layer

All tables in operational_prod.gold_data β€” discover objects outside the current MCP prefix

Browse Other Layers

Other schemas in operational_prod

πŸ” Query Debug Log

No queries yet β€” send a message to see debug info.