Magic Insights MCP server

Connect any MCP client to your Magic Insights workspace and chat with your customer-feedback studies — list studies, search and filter feedback, and read AI-generated themes, topics, suggestions, tags, and sentiment trends.

This is a remote, hosted MCP server. There's nothing to install — point your client at the URL below and authenticate with a workspace API key.

Server URL: https://app.magicinsights.ai/api/mcp
Transport: Streamable HTTP
Auth: API key (bearer token)
Official MCP registry: ai.magicinsights/insights

Get an API key

Create one in the app at Settings → API keys (https://app.magicinsights.ai → your workspace → Settings → API keys), then send it as the Authorization: Bearer <key> header.

Connect

Claude Code

claude mcp add --transport http magic-insights https://app.magicinsights.ai/api/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Cursor

Add to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project), then enable the server in Cursor settings:

{
  "mcpServers": {
    "magic-insights": {
      "url": "https://app.magicinsights.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Other clients (Claude Desktop, Windsurf, MCP Inspector, …)

Most clients use the same mcpServers block above — check your client's docs for where its config file lives. The Claude API MCP connector accepts the same URL with an authorization_token.

Tools

All tools are read-only and scoped to the workspace that owns the API key.

Tool	What it does
`get_workspace`	One-call overview: plan, study count, connected platforms, tags, total feedback, date range, overall sentiment.
`list_studies`	List all studies with description and date range.
`get_study`	A study's summary, generated report, and top themes (with sentiment, ids, and example feedback).
`list_topics`	A study's topics and the themes grouped under each.
`get_theme_feedback`	All feedback under one theme, most relevant first.
`list_suggestions`	A study's AI-synthesized recommended actions (problem, solution, priority).
`get_suggestion_feedback`	The individual feedback behind a single suggestion.
`search_feedback`	Search and filter feedback by keyword, sentiment, platform, tag, or date range.
`get_feedback`	One feedback item in full — content, translation, AI summary, and source metadata.
`get_stats`	Aggregate stats: total count, sentiment breakdown, date range.
`get_trends`	Feedback volume and sentiment over time, bucketed by week or month.
`list_tags`	The workspace's custom tags and how many items carry each.
`list_platforms`	The source platforms feeding the workspace, each with its feedback count.

About

Magic Insights is an AI-driven customer-feedback analytics platform. Learn more at magicinsights.ai.

The server's metadata is published in the official MCP registry; this repository hosts the public server.json and connection docs. The application source is closed.