Skip to content

What is SHINE's MCP server?

An overview of SHINE AI's Model Context Protocol server — what it exposes, who can connect, and how it fits with Claude and ChatGPT.

In one sentence

SHINE AI exposes a Model Context Protocol (MCP) server at https://mcp.shineai.io/v1/mcp that lets Claude, ChatGPT, and any MCP-compatible AI assistant answer questions grounded in your SHINE AI organization’s indexed data.

What MCP is

MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools. It’s maintained by Anthropic and supported by most modern AI apps. Think of it as “USB for AI assistants” — any app that speaks MCP can plug into any server that speaks MCP.

What SHINE’s MCP server exposes

Four tools, grouped into two concerns.

Retrieval (grounded Q&A over your data):

  • shine_ask — ask a question in natural language, get a grounded answer with citations
  • shine_search — raw hybrid retrieval, returns matching chunks grouped by file

ChatGPT connector contract (same substance, ChatGPT-expected names and shapes):

  • search — thin alias over shine_search with ChatGPT’s expected output shape
  • fetch — fetch a specific file by id

Who can connect

Anyone with a SHINE AI account. Sign-in happens via OAuth 2.1 — see the OAuth flow. Access is scoped to your organization; a key issued through sign-in can only see your organization’s data.

What SHINE sees

Only the specific question your assistant forwards to the MCP server, and the data we send back. We don’t see the rest of your conversation with the AI.

Type to search…

↑↓ navigate↵ selectEsc close