Arxivsub Mcp

io.github.arxivsub由 GitHub 账号 arxivsub 发布v0.1.1更新于 Oct 24, 2026

Search arXiv and major AI/ML/CV/NLP/robotics conference papers and their related work.

概览

AI 生成的概览

面向 arXivSub 的 MCP 服务器,让助手检索 arXiv 及主要 AI/ML/CV/NLP/机器人会议论文、查找相关文献并管理个人论文库。

可以做什么
按语义和关键词检索 arXiv 及主要会议论文,可按会议、时效、年份和是否有代码筛选,返回六部分结构化摘要、作者、PDF 与代码链接。为已找到的论文查找最近邻的相关工作,可串联成小型文献综述。运行趋势、排名和细分等研究分析。将论文保存或更新到个人论文库,含笔记、阅读状态和收藏夹,并列出已保存论文与收藏夹。
适用场景
适合让助手在 arXiv 和顶级 AI、CV、ML、NLP、机器人会议上做文献检索、顺着论文查找相关工作,或在不离开聊天客户端的情况下维护阅读库。
运行要求
通过 npx 以 stdio 在本地运行,需要 Node.js;也可在 claude.ai 上使用托管远程连接器。需要 arXivSub 账号,并通过环境变量 ARXIVSUB_SKILL_KEY 提供 API 密钥。需要访问 arXivSub 网关的网络连接。
安装前请注意
服务器持有你的 arXivSub API 密钥,并通过 HTTPS 将查询发送至 arXivSub 网关。该密钥对应付费服务:7 天免费试用,之后属于 arXivSub Pro。检索、相似论文和分析调用会消耗每日配额单位,论文库调用免费。论文库保存与更新工具会写入你的个人论文库。远程连接器会把密钥放在 URL 中。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Arxivsub Mcp,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

arxivsub-mcp

An MCP server for arXivSub — search arXiv and the top AI / CV / ML / robotics / NLP conferences, follow any paper to its related work, build research analytics, and manage a personal library, straight from your AI agent (Claude Desktop, Cursor, Cline, Claude Code, Windsurf, …).

arXivSub summarizes every paper into a structured 6-part breakdown (what it's about, innovations, techniques, datasets, results, limitations) and indexes it for semantic search across arXiv + CVPR, ICCV, ECCV, ICLR, ICML, NeurIPS, AAAI, MICCAI, CoRL, RSS, ACL, EMNLP, ICRA, IROS, IJCAI, SIGGRAPH, SIGGRAPH Asia. This server puts that engine directly in your agent's hands.

Get a free key (7-day trial, no credit card)

  1. Sign in at https://arxivsub.comfyai.app/skills with email or Google.
  2. Copy your API key from the Skills page.
  3. Put it in the server config below as ARXIVSUB_SKILL_KEY.

New accounts get a 7-day free trial of the full skill. After that it's part of arXivSub Pro.

Install

The server runs via npx — no manual install needed. Add it to your MCP client's config.

Claude Desktop (claude_desktop_config.json), Cursor (~/.cursor/mcp.json), Windsurf, Cline, etc.:

json
{  "mcpServers": {    "arxivsub": {      "command": "npx",      "args": ["-y", "arxivsub-mcp"],      "env": { "ARXIVSUB_SKILL_KEY": "your_key_here" }    }  }}

Claude Code:

bash
claude mcp add arxivsub --env ARXIVSUB_SKILL_KEY=your_key_here -- npx -y arxivsub-mcp

Codex CLI:

bash
codex mcp add arxivsub --env ARXIVSUB_SKILL_KEY=your_key_here -- npx -y arxivsub-mcp

Or add it to ~/.codex/config.toml directly:

toml
[mcp_servers.arxivsub]command = "npx"args = ["-y", "arxivsub-mcp"]env = { ARXIVSUB_SKILL_KEY = "your_key_here" }

Restart your client and the tools below become available.

claude.ai (remote connector)

On claude.ai there's nothing to install, add a hosted connector:

  1. Settings → Connectors → Add custom connector

  2. Name: arxivsub

  3. Remote MCP server URL (replace YOUR_KEY with your key from https://arxivsub.comfyai.app/skills):

    https://qtevnmgyobilaanrzidq.supabase.co/functions/v1/mcp?key=YOUR_KEY
  4. Leave the OAuth fields empty and click Add.

This is a hosted Streamable HTTP MCP endpoint, a thin adapter over the arXivSub gateway exposing the same tools. Source: supabase/functions/mcp/index.ts.

Tools

ToolWhat it does
search_papersSemantic + keyword search over arXiv and conferences. Filters: venues, recency, year, has_code. Returns each paper's 6-part summary, authors, PDF, and code link.
find_similar_papersNearest-neighbour "related work" for a paper you already found. Chainable for a mini literature review.
research_insightPivot analytics (the website's Custom Chart engine): trends, rankings, breakdowns. Pro / trial.
library_saveSave/update a paper in your personal library — notes, reading status, collections.
library_listList your saved papers and collections.

Just ask your agent naturally — "find recent papers on LLM safety with code", "more work like this one", "how has interest in diffusion models trended by month", "save this to my "to read" collection" — and it will pick the right tool.

Notes

  • Case-sensitive venues: arxiv, CVPR, ICCV, ECCV, ICLR, ICML, NeurIPS, AAAI, MICCAI, CoRL, RSS, ACL, EMNLP, ICRA, IROS, IJCAI, SIGGRAPH, SIGGRAPH Asia.
  • Quota: each search / similar / insight call uses one daily quota unit; library calls are free. The remaining quota is returned with every result.
  • Privacy / security: the server only holds your skill key and talks to the arXivSub gateway over HTTPS. It has no direct database access and no service credentials.

Development

bash
npm installnpm run build           # tsc -> dist/ARXIVSUB_SKILL_KEY=... node dist/index.js   # run over stdio

License

MIT

来源:README.md,提交 b07c8d4

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

1
  1. v0.1.1最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分