Model Context Protocol (MCP)

Meridian MCP integration - Connect any MCP-compatible AI agent to your company's knowledge base.

Overview

The Meridian Model Context Protocol (MCP) server enables any MCP-compatible AI agent to search and access your company's institutional knowledge directly within conversations.

Works seamlessly with Claude Desktop, Cursor, and any other MCP-compatible client - whether it's a coding assistant, general-purpose AI, or specialized agent.

Ask questions about your internal documents, Slack conversations, emails, and other connected data sources - all without leaving your workflow.

Quick Start

1. Get Your API Key

Sign up at dashboard.trymeridian.dev to get your API key (starts with kt_)

2. Configuration

For Claude Desktop

Add Meridian to your Claude Desktop config file:

{
  "mcpServers": {
    "meridian-search": {
      "url": "https://dashboard.trymeridian.dev/mcp",
      "transport": "http",
      "headers": {
        "X-API-Key": "kt_your_meridian_api_key_here"
      }
    }
  }
}

For Cursor

Add Meridian to your Cursor MCP config:

{
  "mcpServers": {
    "meridian": {
      "url": "https://dashboard.trymeridian.dev/mcp",
      "transport": "http",
      "headers": {
        "X-API-Key": "kt_your_meridian_api_key_here"
      }
    }
  }
}

3. Usage

Restart your MCP client (Claude Desktop, Cursor, etc.) and start asking questions about your company knowledge:

"What did we decide about the pricing strategy?"

"Find the Q4 sales deck"

"Show me recent discussions about the API redesign"

Features

  • Natural Language Search: Ask questions in plain English across all connected data sources
  • Source Citations: Every answer includes links back to original documents
  • Real-time Access: Search your company's knowledge base in real-time
  • Context Preservation: Maintains conversation context for follow-up questions
  • Multi-source: Search across Slack, Email, Drive, Notion, and more simultaneously

Available MCP Tools

meridian_search

Search across all connected company data sources

Parameters:

  • query - Natural language search query
  • max_results - Maximum number of results (default: 10)

meridian_get_context

Retrieve full document content by ID

Parameters:

  • content_id - Document identifier from search results

Coming Soon

Full MCP documentation is under development. Check back soon for detailed setup guides, advanced configuration options, troubleshooting tips, and example workflows.

Questions? Reach out to kn@trymeridian.dev