We built a Cursor integration for running an evidence-grounded first research pass without leaving the editor.
Typical workflow:
- Define a target audience from a plain-language brief.
- Create a panel around a concrete product or marketing decision.
- Ask a focused question or plan a multi-question study.
- Inspect segment disagreement and minority views.
- Export the result as PDF, CSV, XLSX, JSON, or Markdown.
Add the remote server in Cursor Settings → MCP:
{
"mcpServers": {
"minds": {
"type": "http",
"url": "https://getminds.ai/mcp"
}
}
}
OAuth is preferred; scoped Minds API keys also work where secure custom headers are supported. The public repository contains the Cursor plugin manifest and the shared documentation source:
Setup and research-scope notes:
This is decision support, not a substitute for representative human fieldwork. I would welcome feedback on the tool boundaries and which research workflow is most useful inside Cursor.