PaperMCP
# PaperMCP
<div align="center">
[](https://www.npmjs.com/package/@telagod/papermcp)
[](LICENSE)
[](https://www.typescriptlang.org/)
[](https://modelcontextprotocol.io)
**Search and access academic papers from 23+ sources through Model Context Protocol**
[English](README.md) | [简体中文](README.zh-CN.md)
</div>
---
## ❌ Without PaperMCP
- Manually visiting multiple academic databases
- Copy-pasting DOIs and paper IDs between tools
- Switching contexts to download PDFs
- No unified interface for paper search
## ✅ With PaperMCP
- **Single interface** for 23+ academic sources
- **Direct access** from Claude and other AI assistants
- **Automatic PDF download** and text extraction
- **Plugin system** for sensitive sources
---
## 🛠️ Installation
<details>
<summary><b>Install in Claude Desktop</b></summary>
Add to your Claude Desktop config file:
**macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
**Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
#### Using npx (Recommended)
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
#### Using npm global install
```bash
npm install -g @telagod/papermcp
```
```json
{
"mcpServers": {
"papermcp": {
"command": "papermcp"
}
}
}
```
#### With API keys
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"],
"env": {
"SEMANTIC_SCHOLAR_API_KEY": "YOUR_KEY",
"WOS_API_KEY": "YOUR_KEY",
"SCOPUS_API_KEY": "YOUR_KEY"
}
}
}
}
```
</details>
<details>
<summary><b>Install in Cline</b></summary>
Add to Cline MCP settings:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Zed</b></summary>
Add to `~/.config/zed/settings.json`:
```json
{
"context_servers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Continue</b></summary>
Add to Continue config:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Sourcegraph Cody</b></summary>
Add to Cody settings:
```json
{
"cody.experimental.mcp": {
"servers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
}
```
</details>
<details>
<summary><b>Install in Cursor</b></summary>
Add to Cursor MCP settings:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Windsurf</b></summary>
Add to Windsurf config:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Roo Cline</b></summary>
Add to Roo Cline settings:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in OpenHands</b></summary>
Add to OpenHands config:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Void</b></summary>
Add to Void settings:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Claude Code</b></summary>
Use the `claude mcp add` command:
```bash
claude mcp add papermcp -- npx -y @telagod/papermcp
```
Or manually add to `~/.claude.json`:
```json
{
"projects": {
"/your/project/path": {
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
}
}
```
</details>
<details>
<summary><b>Install in Gemini CLI</b></summary>
Add to `~/.config/gemini-cli/settings.json`:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
<details>
<summary><b>Install in Codex CLI</b></summary>
Add to `~/.codex/config.toml`:
```toml
[mcp_servers.papermcp]
command = "npx"
args = ["-y", "@telagod/papermcp"]
```
</details>
<details>
<summary><b>Install in OpenCode</b></summary>
Add to `.opencode.json`:
```json
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"papermcp": {
"type": "local",
"command": ["npx", "-y", "@telagod/papermcp"],
"enabled": true
}
}
}
```
</details>
<details>
<summary><b>Install in Crush CLI</b></summary>
Add to Crush config:
```json
{
"mcpServers": {
"papermcp": {
"command": "npx",
"args": ["-y", "@telagod/papermcp"]
}
}
}
```
</details>
---
## 📚 Supported Platforms
### Core Platforms (16)
| Platform | ID | API Key | Features |
|----------|----|---------|---------|
| arXiv | `arxiv` | ❌ | Search, Download, Read |
| PubMed | `pubmed` | ❌ | Search, Lookup |
| PubMed Central | `pmc` | ❌ | Search, Download, Read |
| bioRxiv | `biorxiv` | ❌ | Search |
| medRxiv | `medrxiv` | ❌ | Search |
| Google Scholar | `google-scholar` | ❌ | Search |
| IACR ePrint | `iacr` | ❌ | Search, Download |
| Semantic Scholar | `semantic` | ⚠️ | Search, Lookup |
| CrossRef | `crossref` | ❌ | Search, Lookup |
| ACM Digital Library | `acm` | ❌ | Search |
| Web of Science | `wos` | ✅ | Search |
| Scopus | `scopus` | ✅ | Search |
| JSTOR | `jstor` | ❌ | Search |
| ResearchGate | `researchgate` | ❌ | Search |
| CORE | `core` | ✅ | Search |
| Microsoft Academic | `microsoft-academic` | ✅ | Search |
### Optional Plugins (7)
Enable via environment variables:
```bash
PLUGIN_SCI_HUB=true
PLUGIN_LIBGEN=true
PLUGIN_UNPAYWALL=true
PLUGIN_OA_BUTTON=true
PLUGIN_SCIENCE_DIRECT=true
PLUGIN_SPRINGER_LINK=true
PLUGIN_IEEE_XPLORE=true
```
> [!WARNING]
> Sci-Hub and LibGen plugins are disabled by default. Use responsibly and comply with local regulations.
---
## 🚀 Usage
### Available Tools (4 Unified Tools)
#### `recommend_platforms`
Get platform recommendations based on field
```typescript
{
"query": "transformer neural network",
"field": "computer-science" // biomedical, physics, mathematics, cryptography, open-access, general
}
```
#### `search_papers`
Search for papers on specified platform
```typescript
{
"platform": "arxiv", // Use recommend_platforms to get suggestions
"query": "machine learning",
"limit": 10
}
```
#### `download_paper`
Download paper PDF from platform
```typescript
{
"platform": "arxiv",
"id": "2301.00001",
"dir": "/path/to/save"
}
```
#### `read_paper`
Extract text from paper PDF
```typescript
{
"platform": "pmc",
"id": "PMC8123456",
"dir": "/path/to/pdfs"
}
```
---
## ⚙️ Configuration
### Environment Variables
```bash
# Optional API keys for enhanced features
SEMANTIC_SCHOLAR_API_KEY=your-key
WOS_API_KEY=your-key
SCOPUS_API_KEY=your-key
CORE_API_KEY=your-key
# Plugin toggles (default: false)
PLUGIN_SCI_HUB=true # No config needed
PLUGIN_LIBGEN=true # No config needed
PLUGIN_UNPAYWALL=true # Requires UNPAYWALL_EMAIL
PLUGIN_OA_BUTTON=true # Optional OA_BUTTON_API_KEY
PLUGIN_SCIENCE_DIRECT=true # Requires ELSEVIER_API_KEY
PLUGIN_SPRINGER_LINK=true # No config needed
PLUGIN_IEEE_XPLORE=true # No config needed
# Plugin API keys
UNPAYWALL_EMAIL=your@email.com # Required for Unpaywall
ELSEVIER_API_KEY=your-key # Required for ScienceDirect
OA_BUTTON_API_KEY=your-key # Optional for Open Access Button
```
### Getting API Keys
| Platform | Link | Notes |
|----------|------|-------|
| **Semantic Scholar** | [Apply](https://www.semanticscholar.org/product/api) | Free, increases rate limit |
| **Web of Science** | [Apply](https://developer.clarivate.com/) | Requires institutional subscription |
| **Scopus** | [Apply](https://dev.elsevier.com/) | Requires institutional subscription |
| **CORE** | [Apply](https://core.ac.uk/services/api) | Free |
| **Unpaywall** | Any email | No registration needed |
| **ScienceDirect** | [Apply](https://dev.elsevier.com/) | Requires institutional subscription |
| **Open Access Button** | [Apply](https://openaccessbutton.org/account) | Optional, free |
---
## 🔧 Development
### Install from source
```bash
git clone https://github.com/telagod/papermcp.git
cd papermcp/ts
npm install
npm run build
npm run dev
```
### Project Structure
```
ts/
├── src/
│ ├── core/ # Core types and config
│ ├── platforms/ # Platform adapters
│ ├── plugins/ # Optional plugins
│ ├── services/ # Registry and tools
│ ├── server/ # MCP server
│ └── utils/ # HTTP, logging
└── dist/ # Compiled output
```
### Adding a Platform
```typescript
import { BasePlatformAdapter } from './baseAdapter.js';
import { addAdapterFactory } from './index.js';
class MyAdapter extends BasePlatformAdapter {
constructor() {
super('my-platform');
}
async search(query: SearchQuery): Promise<SearchResult> {
// Implementation
}
}
addAdapterFactory(() => new MyAdapter());
```
---
## 📄 License
MIT License - see [LICENSE](LICENSE) for details.
---
## 🙏 Acknowledgments
Built with [Model Context Protocol](https://modelcontextprotocol.io) and [MCP TypeScript SDK](https://github.com/modelcontextprotocol/typescript-sdk)
---
<div align="center">
**[⬆ back to top](#papermcp)**
Made with ❤️ for researchers
</div>
TDQS
Scored across 4 tools
Each tool has a distinct role in the workflow: recommend_platforms suggests sources, search_papers finds papers, download_paper retrieves files, and read_paper extracts content. There is no overlap or confusion between these tools.
All tool names follow the verb_noun pattern with consistent snake_case formatting (recommend_platforms, search_papers, download_paper, read_paper). The naming is uniform and predictable.
With only 4 tools, the set is tightly scoped and each tool addresses a necessary step in the paper search and reading pipeline. This is an appropriate size for the server's purpose.
The tool set covers the complete workflow from platform recommendation, to search, download, and reading. There are no obvious gaps in the core functionality a user would expect from a paper management server.