List your AI tool on AI Orion for free. Submit a tool

alphaXiv

Follow the latest research

About alphaXiv

alphaXiv is a platform for discovering, reading and discussing arXiv papers. It connects papers, researchers and organisations and grounds its answers in the underlying work. The home feed lists trending papers with a one-sentence takeaway, contributing institutions, authors, votes and view counts, alongside suggested researchers to follow. It serves researchers who want to track what is worth reading and who is working on a subject, and developers building AI agents that need research access.

Beyond the browser app, alphaXiv runs a Model Context Protocol server at api.alphaxiv.org that gives AI applications, agents and workflows programmatic access to the corpus. The server uses streamable HTTP and advertises OAuth 2.1, so an MCP client opens a browser sign-in on first use; scripts and headless agents can instead send an API key as a bearer token. Native clients such as Claude Code, Claude Desktop, Cursor, VS Code and Zed are supported.

The server exposes 21 tools in three groups. Research tools read the For You and Hot feeds, run discover_papers, an agentic retrieval loop combining keyword and embedding search that returns 5 to 15 ranked papers, fetch a paper as an AI-generated report or full text, return the PDF pages relevant to a set of queries so citations can be built, and read files from a paper's GitHub repository. Researcher tools find people by subject, organisation, career history or coauthorship.

Feed retrieval makes no model calls and does not count against the assistant quota, while search and analysis tools do. Browser-hosted MCP integrations inside claude.ai or chatgpt.com are not supported because CORS is restricted to first-party origins; a local bridge such as mcp-remote is needed for browser-based assistants. Researchers can sign up, publish, follow authors and claim their own profile, and library management is also available through the MCP tools.

Use cases

  • Literature discovery and related-work searches
  • Exploring a paper's codebase and managing a reading library
  • Finding who works on a subject or an organisation's roster
  • Building citations from the relevant pages of a PDF
  • Keeping up with trending research

Key features

  • Explore feed of trending arXiv papers with discussion and activity
  • Follow authors and researchers
  • Grounded literature review
  • MCP server giving AI agents programmatic access to the research corpus
  • discover_papers: agentic retrieval combining keyword and embedding search
  • Paper content as an AI-generated report or full extracted text
  • Page-level PDF answers for building citations
  • Reading files from a paper's GitHub repository
  • Researcher search by subject, organisation, career history or coauthorship
  • For You and Hot feeds available to agents
  • OAuth 2.1 sign-in or API key authentication

Pros and cons

Pros

Feed retrieval does not use the assistant quota
OAuth sign-in needs no configuration

Cons

Browser-hosted MCP integrations (claude.ai, chatgpt.com) are not supported
Search and analysis tools count against an assistant quota

Reviews

Loading reviews…

Frequently asked questions

Which MCP clients are supported?
Native MCP clients only: Claude Code, Claude Desktop, Cursor, VS Code, Zed and CLI bridges such as mcp-remote.
How do scripts and headless agents authenticate?
By creating an API key under Settings > API Keys and sending it as a bearer token, which skips the OAuth flow.

Alternatives to alphaXiv

More in AI Research Tool