arXiv2MD

Category: Agent · Official site: https://www.arxiv2md.org/

Clean paper HTML into LLM-friendly Markdown.

Paper HTML often carries noisy tags, scripts, and messy markup that consume context windows unnecessarily. This utility is designed for developers and researchers who ingest HTML versions of research papers into language model applications. Choose it over plain text extractors or LaTeX parsers when your source material is web-based HTML and you want clean Markdown headers, lists, and formatting without maintaining raw markup or dealing with source code archives.

It belongs to the agent tooling step: feeding papers to software — agents, pipelines and scripts — in clean, structured form.

On PaperBridge you do not need to rebuild this link by hand for every paper: the app substitutes the arXiv ID (and, where a service needs it, the title or DOI) into the service's own URL pattern. Links to the tool's homepage — this service offers no stable per-paper deep link.

Open with an example paper Try it on PaperBridge

Example paper: Attention Is All You Need (arXiv:1706.03762). The first button opens arXiv2MD for that paper (this service takes you to its home page); the second opens the paper's PaperBridge page, where every tool is one click away.

Related workflow

This tool is one stop on the route from a paper to its code, datasets, and benchmarks — the guide gives the full sequence and the checks to run before you trust a repository. Read: How to Find Code for Any arXiv Paper.

More Agent tools

arXiv TXT — Turn a paper into plain text an LLM can ingest.
ArcXiv — Connect search APIs, structured Markdown, and MCP.
MarkXiv — Generate Markdown from LaTeX structure.
arxiv-mcp — Search, download, and read arXiv by section through MCP.
arxiv.py — Search and download papers programmatically with Python.
Scry — Bring arXiv, Reddit, and HN into an agent.
alphaxiv-mcp — Let agents call alphaXiv full text and reports.
DeepXiv — A full-text research data layer for agents.

See the full tool directory or the Agent category page.