PaperDog for AI Agents

Empower AI assistants with access to thousands of research papers through Model Context Protocol

No Authentication Rate Limited Free Service

Quick Start for AI Agents

PaperDog provides MCP tools for AI research paper discovery. Here's how to integrate:

MCP Endpoint:

http://paperdog.org/mcp

Service Discovery:

http://paperdog.org/.well-known/mcp

Available Tools:

paperdog_search_papers - Search arXiv and HuggingFace papers
paperdog_get_daily_papers - Get curated daily papers
paperdog_get_paper_details - Get detailed paper information
paperdog_get_categories - List available research categories
paperdog_get_archive_papers - Access historical archives

AI Agent Capabilities

Smart Search

Search across thousands of papers with advanced filtering by category, relevance score, and date ranges.

AI Analysis

Get AI-powered analysis including challenges, innovations, experiments, and insights for each paper.

Daily Curation

Access daily curated top papers with comprehensive analysis and relevance scoring.

Historical Archive

Query historical paper archives with advanced search and filtering capabilities.

Rate Limited

Fair usage with 100 requests per hour and 1000 requests per day per IP address.

Integration Examples

Claude Integration

Add PaperDog MCP to Claude's configuration:

{
  "mcpServers": {
    "paperdog": {
      "command": "curl",
      "args": ["-X", "POST", "http://paperdog.org/mcp",
               "-H", "Content-Type: application/json",
               "-d", "@-"]
    }
  }
}
                            

Then ask Claude:

"Search for recent papers on transformer architectures and analyze the top 3 results"
                            

ChatGPT Plugin

Create a custom ChatGPT plugin that calls PaperDog's MCP endpoint:

POST http://paperdog.org/mcp
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "paperdog_search_papers",
    "arguments": {
      "query": "generative AI",
      "limit": 10
    }
  }
}
                            

Technical Details

Protocol & Specifications

Protocol

Model Context Protocol (MCP) - JSON-RPC 2.0

Authentication

None required (IP-based rate limiting)

Rate Limits

100 requests per hour
1000 requests per day
Per IP address

Response Format

JSON with paper metadata, analysis, and search results

Error Handling

Standard JSON-RPC error codes with descriptive messages

Cache Strategy

5-minute cache for search results, 15-minute cache for daily papers