Provides tools for searching academic papers on arXiv with automatic URL encoding and semantic similarity analysis using AI-powered embeddings to find the most relevant research papers
LitSynth MCP Server
A Model Context Protocol (MCP) server for intelligent academic paper discovery and semantic search using ArXiv. This server provides tools for searching academic papers and performing semantic similarity analysis using state-of-the-art sentence transformers.
Features
ArXiv Search: Query ArXiv database with automatic URL encoding for complex search terms
Semantic Search: Find papers most relevant to your research using AI-powered semantic similarity
Robust Error Handling: Graceful handling of network issues and malformed data
Flexible Input: Support for various query formats including spaces and special characters
Tools Available
1. greet(name: str)
Simple greeting function for testing server connectivity.
Parameters:
name
: String - Name to greet
Returns: Greeting message
2. search_query_arxiv(query: str, max_results: int = 5)
Search ArXiv database for academic papers matching your query.
Parameters:
query
: String - Search terms (automatically URL encoded)max_results
: Integer - Maximum number of results to return (default: 5)
Returns: Structured response with papers including:
Title
Authors
Summary/Abstract
ArXiv link
Status message
Example:
3. search_semantic_arxiv(query: str, papers: list, top_k: int = 5)
Perform semantic search on a list of papers to find the most relevant ones.
Parameters:
query
: String - Research query for semantic matchingpapers
: List - Papers to search through (fromsearch_query_arxiv
or manual list)top_k
: Integer - Number of most relevant papers to return (default: 5)
Returns: Ranked papers with similarity scores including:
Title
Summary
Authors
ArXiv link
Similarity score (0-1)
Example:
4. search_hf_datasets(query: str, limit: int = 5)
Perform a research on Hugging Face to find the most relevant datasets on a topics
Parameters:
query
: String - Research query for matching datasetslimit
: Integer - Number of the limit of datasets showed
Returns: Ranked datasets info
Id
Description
Url
5. get_dataset_details(dataset_id: str)
Get detailed information about a specific Hugging Face dataset. It provides comprehensive analysis including structure, usage examples, and research applications.
Parameters:
dataset_id
: String - The dataset ID from Hugging Face (e.g., "microsoft/COCO")
Returns: Detailed dataset information including metadata, structure, and usage examples
Dataset Id
Url
Basic Info
Structure
Files
Usage Information
Research Applications
6. explore_dataset_files(dataset_id: str)
Explore the structure of a Hugging Face dataset without downloading it entirely. Shows available files/splits and provides a sample preview of the data.
Parameters:
dataset_id
: String - The dataset ID from Hugging Face (e.g., "microsoft/COCO")
Returns: Dataset structure information including files, splits, and sample data
Dataset Id
Url
Structure
Available Files
Sample Preview
Summary
7. explore_dataset_structure(dataset_id: str, split_name: str = None)
Explore the structure of a Hugging Face dataset with detailed split information.
Parameters:
dataset_id
: String - The dataset ID from Hugging Face (e.g., "microsoft/COCO")split_name
: Optional specific split to explore (e.g., "train", "test", "validation")
Returns: Dataset structure overview with splits, file types, and sample data
Dataset Id
Url
Overview
Available Splits
File Types
Sample Data
Description
Next Steps
Installation
Prerequisites
Python 3.8+
pip
Setup
Clone or download the project files
Install dependencies:
Run the MCP server:
Dependencies
The project requires the following packages (see requirements.txt
):
fastmcp>=0.1.0
- MCP frameworkfeedparser>=6.0.10
- RSS/Atom feed parsing for ArXiv APIrequests>=2.31.0
- HTTP requestssentence-transformers>=2.2.2
- Semantic search and embeddingstorch>=2.0.0
- PyTorch for neural networkstransformers>=4.21.0
- Hugging Face transformersnumpy>=1.21.0
- Numerical computing
Project Structure
Usage Examples
Basic ArXiv Search
Search for papers on a specific topic:
Semantic Paper Discovery
Find the most relevant papers from a search result:
Handling Complex Queries
The server automatically handles special characters and spaces:
Technical Details
Semantic Search Model
The server uses the sentence-transformers/all-MiniLM-L6-v2
model for semantic embeddings. This model:
Provides 384-dimensional sentence embeddings
Balances speed and accuracy
Works well for academic text similarity
Error Handling
The server includes comprehensive error handling:
URL Encoding: Automatic handling of spaces and special characters
Network Errors: Graceful degradation when ArXiv is unavailable
Data Validation: Safe handling of missing or malformed paper data
Empty Results: Informative messages when no papers are found
Response Format
All functions return structured responses:
Troubleshooting
Common Issues
"URL can't contain control characters" error:
This is fixed in the current version with automatic URL encoding
Make sure you're using the latest version of the server
"No papers found" result:
Check your query spelling
Try broader search terms
Verify ArXiv service availability
Slow semantic search:
First run downloads the transformer model (~90MB)
Subsequent runs are much faster
Consider reducing
top_k
for faster results
Memory issues:
The sentence transformer model requires ~500MB RAM
Reduce batch sizes if experiencing memory problems
Contributing
Feel free to submit issues, feature requests, or pull requests to improve the AI Research Assistant.
License
This project is open source. Please check individual dependency licenses for commercial use.
Acknowledgments
ArXiv for providing free access to academic papers
Sentence Transformers for semantic search capabilities
FastMCP for the MCP server framework
This server cannot be installed
remote-capable server
The server can be hosted and run remotely because it primarily relies on remote services or has no dependency on the local environment.
Enables intelligent academic paper discovery through ArXiv search and AI-powered semantic similarity analysis. Helps researchers find and rank the most relevant papers using natural language queries and state-of-the-art sentence transformers.
- Features
- Tools Available
- 1. greet(name: str)
- 2. search_query_arxiv(query: str, max_results: int = 5)
- 3. search_semantic_arxiv(query: str, papers: list, top_k: int = 5)
- 4. search_hf_datasets(query: str, limit: int = 5)
- 5. get_dataset_details(dataset_id: str)
- 6. explore_dataset_files(dataset_id: str)
- 7. explore_dataset_structure(dataset_id: str, split_name: str = None)
- Installation
- Dependencies
- Project Structure
- Usage Examples
- Technical Details
- Troubleshooting
- Contributing
- License
- Acknowledgments