---
title: "MCP"
description: "Connect AI agents and developer tools to ScholarXIV through the Model Context Protocol. 33 tools for paper search, collections, bookmarks, comments, and research."
canonical_url: "https://scholarxiv.com/developers/docs/mcp"
markdown_url: "https://scholarxiv.com/developers/docs/mcp.md"
---

# MCP
URL: /developers/docs/mcp
LLM index: /llms.txt
Description: Connect AI agents and developer tools to ScholarXIV through the Model Context Protocol. 33 tools for paper search, collections, bookmarks, comments, and research.
Related: mcp/quickstart, mcp/connect, mcp/tools

# ScholarXIV MCP

The ScholarXIV MCP server lets AI agents and developer tools interact with ScholarXIV through the [Model Context Protocol](https://modelcontextprotocol.io). It provides 33 tools covering paper search, collections, bookmarks, comments, AI research chat, and more.

## MCP vs Papers API

| Feature | Papers API | MCP |
|---------|------------|-----|
| **Protocol** | REST (HTTP) | Model Context Protocol (Streamable HTTP) |
| **Endpoint** | `GET/POST /api/v1/papers/search` | `POST /api/mcp` |
| **Tools** | 1 endpoint (paper search) | 33 tools (search, collections, bookmarks, comments, chat, account) |
| **Auth** | Same `sxv_` API key | Same `sxv_` API key |
| **Best for** | Direct integration into apps | AI agents, IDE assistants, research workflows |

Use the **Papers API** if you need programmatic paper search in a traditional app. Use **MCP** if you want an AI agent to autonomously search papers, build collections, bookmark findings, and discuss research.

## What You Can Do

- **Search** — find papers by title, author, abstract, category, or keyword
- **Read** — get full paper metadata and abstracts
- **Organize** — create collections, add/remove papers, manage members
- **Interact** — bookmark, like, comment on papers
- **Research** — ask AI research questions with paper context
- **Discover** — browse public collections from other researchers

## Prerequisites

1. A ScholarXIV account
2. An API key (create one at [/developers/dashboard](/developers/dashboard))
3. An MCP-compatible client (Claude Desktop, Cursor, VS Code, etc.)

## Quick Example

Add this to your MCP client config:

```json
{
  "mcp": {
    "scholarxiv": {
      "type": "remote",
      "url": "https://www.scholarxiv.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sxv_your_api_key_here"
      }
    }
  }
}
```

Then ask your AI agent: *"Search for recent papers on protein folding and bookmark the top 3."*

## Next Steps

- [Quick Start](/developers/docs/mcp/quickstart) — go from zero to first tool call in 5 minutes
- [How to Connect](/developers/docs/mcp/connect) — endpoint, auth, and transport details
- [Client Configuration](/developers/docs/mcp/clients) — config examples for every major client
- [Tool Catalog](/developers/docs/mcp/tools) — all 33 tools with parameters and examples
- [Usage Examples](/developers/docs/mcp/examples) — multi-step research workflows

> [!NOTE]
> MCP calls share the same rate limit as the REST API. Your plan's hourly limit applies to both.

## Sitemap

See the full [sitemap](/sitemap.md) for all pages.
Docs-scoped sitemap: [/docs/sitemap.md](/docs/sitemap.md).
Well-known sitemap: [/.well-known/sitemap.md](/.well-known/sitemap.md).