# arXivv

> The missing data layer for arXiv, built for AI and LLMs.

arXivv serves every arXiv paper as Markdown, JSON and plain text: append an extension to a paper ID, e.g. `https://arxivv.org/1706.03762.md`. Swap `arxiv.org` for `arxivv.org` in any existing URL and it keeps working. Agents can also connect to the stateless MCP server at `https://mcp.arxivv.org/mcp` (also reachable at `https://arxivv.org/mcp`).

## Docs

- [Introduction](https://docs.arxivv.org/): arXivv is the missing data layer for arXiv, built for AI and LLMs. Swap the host, or add an extension to a paper ID.
- [Drop-in compatibility](https://docs.arxivv.org/drop-in): Swap arxiv.org for arxivv.org in any URL or client. Every arXiv path and parameter is forwarded to arXiv and its response returned as-is.
- [Formats and versions](https://docs.arxivv.org/formats): Markdown, JSON and plain text for every paper, figure assets, versioned and unversioned URLs, and how responses are cached.
- [Document model](https://docs.arxivv.org/document-model): Every paper is parsed into one canonical document tree. Every node has a stable anchor, and every format is a rendering of the tree.
- [JSON schema](https://docs.arxivv.org/json): The PaperJson, MetadataJson and Outline shapes, a trimmed example for 1706.03762, and the schema 1 compatibility promise.
- [Inferred metadata](https://docs.arxivv.org/inferred): LLM-generated fields per paper, labelled with the models and prompt version that produced them. Status is pending, ready or failed.
- [First request](https://docs.arxivv.org/first-request): How arXivv behaves the first time a paper is requested, and how to wait for inferred fields or build asynchronously with Prefer.
- [MCP server](https://docs.arxivv.org/mcp): A stateless MCP server at https://mcp.arxivv.org/mcp (also https://arxivv.org/mcp) with seven tools for reading papers by anchor.
- [Errors and limits](https://docs.arxivv.org/errors-and-limits): Problem Details errors with stable codes, the per-IP rate limit, and the headers that come with them.
- [Changelog](https://docs.arxivv.org/changelog): Changes to arXivv's routes, formats, schema, MCP tools and limits.

## arXivv API

- [arXivv API](https://docs.arxivv.org/api): Every arXiv paper as Markdown, JSON and plain text, plus its parts by anchor. One rule: /<paper>.<format>. Any other path is proxied to arXiv unchanged…

### Papers

- [Paper as Markdown](https://docs.arxivv.org/api/papers/get-paper-markdown): The whole paper rendered from the document tree: YAML front matter, # Title, ## Abstract, body… Reference for the GET /{idv}.md endpoint in the arXivv API.
- [Paper as JSON](https://docs.arxivv.org/api/papers/get-paper-json): Metadata, build info, approximate token count, the full document tree and inferred fields… Reference for the GET /{idv}.json endpoint in the arXivv API.
- [Paper as plain text](https://docs.arxivv.org/api/papers/get-paper-text): The Markdown rendering with formatting stripped; math is kept as LaTeX. Reference for the GET /{idv}.txt endpoint in the arXivv API.

### Parts

- [Paper metadata](https://docs.arxivv.org/api/parts/get-metadata): arXiv metadata and URLs without building the paper. build and inferred are present… Reference for the GET /{idv}/metadata.json endpoint in the arXivv API.
- [Paper outline](https://docs.arxivv.org/api/parts/get-outline): Title, authors, abstract, inferred summary, and the section tree with per-section token… Reference for the GET /{idv}/outline.json endpoint in the arXivv API.
- [One section as Markdown](https://docs.arxivv.org/api/parts/get-section): Markdown for one section subtree, heading included. anchor is a LaTeXML id… Reference for the GET /{idv}/sections/{anchor}.md endpoint in the arXivv API.
- [Bibliography](https://docs.arxivv.org/api/parts/get-references): Every reference with its text, inline content and any arXiv id, DOI or URL. Reference for the GET /{idv}/references.json endpoint in the arXivv API.
- [Figures and tables](https://docs.arxivv.org/api/parts/get-figures): Every figure (captions, absolute image URLs) and table (cells), with the paragraphs… Reference for the GET /{idv}/figures.json endpoint in the arXivv API.
- [One figure](https://docs.arxivv.org/api/parts/get-figure): One figure by LaTeXML id (S3.F1) or number alias (F2). Reference for the GET /{idv}/figures/{anchor}.json endpoint in the arXivv API.
- [Figure image or other HTML asset](https://docs.arxivv.org/api/parts/get-asset): A file referenced by the arXiv HTML (e.g. Figures/ModalNet-21.png), mirrored… Reference for the GET /{idv}/assets/{path} endpoint in the arXivv API.

## Agent skills

- [arxivv](https://docs.arxivv.org/.well-known/agent-skills/arxivv/SKILL.md): Read the arXivv docs instead of answering from memory. The missing data layer for arXiv, built for AI and LLMs. Use when a task involves arXivv: how it works, how to set it up or configure it, or its API.

## Agent resources

- [llms-full.txt](https://docs.arxivv.org/llms-full.txt): The full Markdown of every page in one file.
- [Page Markdown](https://docs.arxivv.org/index.md): Append `.md` to any page URL to fetch that page as raw Markdown.
- [JSON API](https://docs.arxivv.org/api/docs/pages.json): Page index of the JSON docs API; each entry links the page's JSON and Markdown forms. Described by the OpenAPI document at https://docs.arxivv.org/openapi.json.
- [Agent skills](https://docs.arxivv.org/.well-known/agent-skills/index.json): Agent Skills discovery index of the skills this site publishes.
- [API catalog](https://docs.arxivv.org/.well-known/api-catalog): RFC 9727 linkset of the APIs documented here.
- [AI catalog](https://docs.arxivv.org/.well-known/ai-catalog.json): ARD manifest of the agent-facing resources on this site (MCP server, skills, APIs).
- [agent-readability.json](https://docs.arxivv.org/agent-readability.json): Manifest of every agent-facing artifact on this site.
- [Sitemap](https://docs.arxivv.org/sitemap.xml): Every indexable page URL with its last-modified date.