kivv is an automatic arXiv research paper discovery and AI intelligent summarization system, integrated with Claude Desktop via the MCP protocol, supporting multi - user, RSS subscription, and cost optimization, and built on Cloudflare Workers.
2 points
7.0K

What is kivv?

kivv is an automated research assistant designed specifically for researchers, scholars, and students. It automatically searches arXiv (the world's largest pre - print paper library) daily to find the latest papers matching your research interests, and then uses Claude AI to generate concise and understandable summaries. You can use it directly in the Claude Desktop chat interface without manually searching and reading a large number of papers.

How to use kivv?

Using kivv is very simple: 1) Deploy it to Cloudflare Workers (the free tier is available), 2) Configure the MCP server in Claude Desktop, 3) Set your research topics of interest. After that, the system will automatically collect and summarize papers for you. You can subscribe via RSS or query directly in Claude.

Applicable scenarios

• Researchers track the latest progress in specific fields • Postgraduate students look for relevant literature and inspiration • Interdisciplinary researchers understand the dynamics of other fields • Anyone who needs to read academic papers regularly but has limited time • Scenarios where teams collaborate to share research findings

Main features

Daily automated search
The system automatically scans arXiv every day, filters relevant papers according to the topic keywords you set (e.g.,'machine learning', 'quantum computing'), without manual operation.
AI intelligent summary
Use Claude 3.5 Sonnet to generate paper summaries. For complex papers, the Haiku model is first used for pre - screening to save costs and ensure high - quality summaries with controllable costs.
Claude Desktop integration
Integrated directly into Claude Desktop via the Model Context Protocol (MCP). You can directly query papers and get recommendations in the chat interface for a seamless experience.
Multi - user support
Supports multiple users to use independently. Each user can set their own research interests and view their personal paper library, with completely isolated data.
RSS subscription
Generate a dedicated RSS/Atom subscription link for each user, which can be subscribed to using any reader (e.g., Feedly, Inoreader) to follow the latest research like following a blog.
Cost optimization
Intelligently use different Claude models: use Haiku for simple summaries and Sonnet for complex analyses. Run on the Cloudflare free tier, with a monthly cost of about $3 (for 2 users).
Advantages
Fully automated: Set it once and get the latest research automatically every day
Save time: AI summaries allow you to quickly understand the core of papers without reading the full text
Integrated experience: Use it directly in Claude without switching applications
Extremely low cost: Cloudflare free tier + intelligent model selection, almost free for personal use
Flexible subscription: RSS support allows you to receive updates using your preferred tools
Privacy protection: User data is isolated, and only public arXiv papers are processed
Limitations
Only supports arXiv: Currently does not support other paper libraries (e.g., PubMed, IEEE Xplore)
Requires technical deployment: Basic knowledge of Cloudflare and API configuration is required
Depends on the Claude API: A valid Anthropic API key is required
Summaries may miss details: AI summaries cannot completely replace careful reading of the full text
Delayed updates: Batch - processed daily, not real - time push

How to use

Prepare the environment
Register a Cloudflare account and enable Workers, D1, KV, and R2 services. Obtain an Anthropic Claude API key. Install the Bun package manager.
Clone and configure
Download the kivv code, copy the environment configuration file, and fill in your API key and Cloudflare account information.
Deploy to Cloudflare
Use the provided script or manually deploy two Workers: the automation Worker (to handle daily tasks) and the MCP server Worker (to communicate with Claude).
Configure Claude Desktop
Add the MCP server information, including the URL and API key, to the Claude Desktop configuration file.
Set research interests
Set your research topics of interest through the Claude chat interface or API, and the system will start collecting relevant papers for you.

Usage examples

A computer science postgraduate student tracks the latest progress
Alice is a postgraduate student in the field of machine learning. She sets her interests as ['deep learning','reinforcement learning', 'computer vision']. Every morning, she views the summaries of 3 - 5 selected papers pushed by kivv through her RSS reader, quickly understanding the dynamics of the field and saving several hours of manual searching time.
An interdisciplinary researcher explores a new field
Bob is a biologist who wants to understand the application of AI in drug discovery. He sets his interests as ['AI drug discovery', 'bioinformatics', 'protein folding']. kivv helps him filter relevant papers on arXiv, avoiding being overwhelmed by a large number of irrelevant CS papers.
A research team shares findings
A 5 - person research team each has their own kivv account but shares some common interest topics. Before their weekly meetings, they learn about the latest papers through their respective RSS subscriptions and then discuss relevant research in the meetings.

Frequently Asked Questions

Is kivv free?
Do I need programming knowledge to use it?
How is data privacy ensured?
What is the quality of the summaries? Will important information be missed?
Does it support papers in Chinese or other languages?
What should I do if I encounter problems during deployment?

Related resources

Complete deployment guide
Detailed step - by - step deployment instructions, including screenshots and troubleshooting
GitHub repository
Source code, issue feedback, and the latest updates
Cloudflare Workers documentation
Understand the Cloudflare Workers platform
Model Context Protocol
Official specification of the MCP protocol
arXiv official website
Source library of papers

Installation

Copy the following command to your Client for configuration
Note: Your key is sensitive information, do not share it with anyone.

Alternatives

A
Acemcp
Acemcp is an MCP server for codebase indexing and semantic search, supporting automatic incremental indexing, multi-encoding file processing, .gitignore integration, and a Web management interface, helping developers quickly search for and understand code context.
Python
8.1K
5 points
K
Klavis
Klavis AI is an open-source project that provides a simple and easy-to-use MCP (Model Context Protocol) service on Slack, Discord, and Web platforms. It includes various functions such as report generation, YouTube tools, and document conversion, supporting non-technical users and developers to use AI workflows.
TypeScript
13.4K
5 points
M
MCP
The Microsoft official MCP server provides search and access functions for the latest Microsoft technical documentation for AI assistants
11.7K
5 points
S
Scrapling
Scrapling is an adaptive web scraping library that can automatically learn website changes and re - locate elements. It supports multiple scraping methods and AI integration, providing high - performance parsing and a developer - friendly experience.
Python
11.5K
5 points
A
Apple Health MCP
An MCP server for querying Apple Health data via SQL, implemented based on DuckDB for efficient analysis, supporting natural language queries and automatic report generation.
TypeScript
10.5K
4.5 points
A
Annas MCP
The MCP server and CLI tool of Anna's Archive are used to search for and download documents on the platform and support access through an API key.
Go
6.7K
4.5 points
S
Search1api
The Search1API MCP Server is a server based on the Model Context Protocol (MCP), providing search and crawling functions, and supporting multiple search services and tools.
TypeScript
16.4K
4 points
D
Duckduckgo MCP Server
Certified
The DuckDuckGo Search MCP Server provides web search and content scraping services for LLMs such as Claude.
Python
53.8K
4.3 points
M
Markdownify MCP
Markdownify is a multi-functional file conversion service that supports converting multiple formats such as PDFs, images, audio, and web page content into Markdown format.
TypeScript
28.0K
5 points
N
Notion Api MCP
Certified
A Python-based MCP Server that provides advanced to-do list management and content organization functions through the Notion API, enabling seamless integration between AI models and Notion.
Python
17.4K
4.5 points
G
Gitlab MCP Server
Certified
The GitLab MCP server is a project based on the Model Context Protocol that provides a comprehensive toolset for interacting with GitLab accounts, including code review, merge request management, CI/CD configuration, and other functions.
TypeScript
19.0K
4.3 points
D
Duckduckgo MCP Server
Certified
The DuckDuckGo Search MCP Server provides web search and content scraping services for LLMs such as Claude.
Python
53.8K
4.3 points
U
Unity
Certified
UnityMCP is a Unity editor plugin that implements the Model Context Protocol (MCP), providing seamless integration between Unity and AI assistants, including real - time state monitoring, remote command execution, and log functions.
C#
22.4K
5 points
F
Figma Context MCP
Framelink Figma MCP Server is a server that provides access to Figma design data for AI programming tools (such as Cursor). By simplifying the Figma API response, it helps AI more accurately achieve one - click conversion from design to code.
TypeScript
51.2K
4.5 points
G
Gmail MCP Server
A Gmail automatic authentication MCP server designed for Claude Desktop, supporting Gmail management through natural language interaction, including complete functions such as sending emails, label management, and batch operations.
TypeScript
18.1K
4.5 points
M
Minimax MCP Server
The MiniMax Model Context Protocol (MCP) is an official server that supports interaction with powerful text-to-speech, video/image generation APIs, and is suitable for various client tools such as Claude Desktop and Cursor.
Python
36.7K
4.8 points
AIBase
Zhiqi Future, Your AI Solution Think Tank
© 2025AIBase