Skip to content
Back to search

deepxiv_sdk

Talk to research papers like talking to authors - Python package with AI agent for arXiv papers

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/deepxiv--deepxiv-sdk/badge.svg)](/listings/deepxiv--deepxiv-sdk)

HTML

<a href="/listings/deepxiv--deepxiv-sdk"><img src="/api/listings/deepxiv--deepxiv-sdk/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

pip install deepxiv-sdk

`deepxiv fa` (1stAuthor domains, including talent) needs 1.2.0 or later: `pip install -U deepxiv-sdk`.

`deepxiv` auto-registers a token on first use. Agentic commands (`ask`, `fa ask`) need a registered key (300 calls/day free); the auto-registered token gets 3 lifetime trial `fa ask` calls. Get a key at [data.rag.ac.cn/register](https://data.rag.ac.cn/register), then:

deepxiv config --token YOUR_REGISTERED_KEY

Every account gets 300 agentic calls/day free, on a pool separate from the general daily limit.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 248 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <h1 align="center">DeepXiv</h1>
  2. <p align="center"><em>The data layer agentic search is missing — full paper text, real citations, and the people behind them.</em></p>
  3. <p align="center">
  4. <a href="https://1stauthor.com/assistant/">Live system</a> ·
  5. <a href="USAGE.md">Full documentation</a> ·
  6. <a href="https://data.rag.ac.cn/api/docs">API docs</a> ·
  7. <a href="https://data.rag.ac.cn/status">Status</a> ·
  8. <a href="https://arxiv.org/abs/2603.00084"><img src="https://img.shields.io/badge/arXiv-2603.00084-b31b1b" alt="arXiv"></a> ·
  9. <a href="README.zh.md">中文</a>
  10. </p>
  11. <p align="center">
  12. <img src="./assets/deepxiv_cover.jpg" width="100%" alt="DeepXiv — the best vertical-domain data service for agents">
  13. </p>
  14. > [!IMPORTANT]
  15. > **Capacity: about 50 requests per second, shared by everyone.** DeepXiv's paper service and the 1stAuthor services run on small servers with limited compute and bandwidth. Donations to keep it running are very welcome 🙏. If you need large volumes, email `tommy[at]chien.io` and we'll work out a plan together.
  16. ---
  17. ## What it is
  18. A CLI and Python SDK over a service that has already read the literature. Papers arrive parsed into sections rather than PDFs, retrieval runs over full bodies rather than abstracts, and an agentic endpoint answers questions with citations that resolve to real arXiv IDs and URLs.
  19. Four things it does:
  20. | | Command | |
  21. |---|---|---|
  22. | **Ask the literature** | `deepxiv ask` | A question in, an answer out, cited with real arXiv IDs |
  23. | **Ask the web** | `deepxiv ask --web` | Same, over Google plus cached page bodies |
  24. | **Read a paper in layers** | `deepxiv search` / `paper` | Search, judge, then read only the section you need |
  25. | **Find the people** | `deepxiv fa … talent` | Who works on a topic, where, and what their record is |
  26. | **Watch US stocks** 🆕 | `deepxiv fa … stocks` | 8,000+ tickers: prices, per-ticker news, filings, financials, macro — point-in-time, and `fa ask -d stocks` for a cited analysis |
  27. | **Go beyond papers** | `deepxiv fa` | Statutes, judgments, trials, drugs, grants, filings, standards, CVEs — via 1stAuthor |
  28. ## What problem it solves
  29. An agent researching a topic has bad options. Search APIs return ten blue links and abstracts — enough to name a paper, never enough to answer "what speedup does it report on HumanEval". PDFs answer that, but cost 50k tokens each and arrive as a wall of text with no structure to navigate.
  30. DeepXiv removes the tradeoff. Papers are pre-parsed, so an agent can spend 300 tokens on a TLDR to decide whether to spend 5k on the Methods section. Questions that need evidence go to the agentic endpoint, which reads source text and hands back an answer with citations you can check. And because knowing *who* does the work is half of research, the same token searches scholars — plus law, clinical trials, drugs, grants and more through [1stAuthor](https://1stauthor.com)'s domains.
  31. ## DeepXiv × 1stAuthor
  32. DeepXiv works closely with [1stAuthor](https://1stauthor.com) to serve data beyond papers: researchers, US stocks, statutes, court judgments, clinical trials, drugs, grants, filings, standards and CVEs, all reached through `deepxiv fa` with the same token. **→ [The `deepxiv fa` guide](FA.md)**
  33. One CLI, fifteen domains, the same three verbs: `search` → `read` → `ask`.
  34. | Domain | What's in it |
  35. |---|---|
  36. | **`stocks`** 🆕 | **8,000+ US tickers: prices, per-ticker news and events, SEC filings, financials, macro series and news — all point-in-time** |
  37. | `talent` | AI researcher profiles: education, career, top papers, metrics |
  38. | `appraise` | What academicians and fellows said when they cited a paper |
  39. | `law` · `us_law` · `uk_eu_law` · `fr` | Chinese and 25 other jurisdictions' statutes; US statutes and regulations; UK and EU legislation; Federal Register |
  40. | `cases` | Chinese and US court judgments |
  41. | `trials` · `drugs` · `grants` | ClinicalTrials.gov, openFDA drug labels, NSF / NIH grants |
  42. | `filings` · `standards` | SEC and Chinese company filings, Chinese national standards |
  43. | `cve` · `osv` | NVD + CISA KEV + EPSS; open-source package advisories and malicious-package reports |
  44. > **Pass `--domain` (`-d`).** Auto-routing (`fa ask` without `-d`) is still in early development and can pick the wrong domain. Name the domain yourself; `deepxiv fa domains` lists them and `deepxiv fa spec DOMAIN` shows its filters.
  45. ```bash
  46. # People: China-based students working on world models
  47. deepxiv fa search talent "world model" -F career_stage=student -F location=北京,上海 -k 20
  48. # Which academicians / fellows cited DeepSeek-R1, and what the positive ones said
  49. deepxiv fa read appraise arxiv:2501.12948 -l brief
  50. deepxiv fa read appraise arxiv:2501.12948 -p stance=positive -p org=CAS,CAE
  51. # Law and cases: statutes plus US Supreme Court judgments
  52. deepxiv fa search law "个人信息保护" -k 5
  53. deepxiv fa search cases "qualified immunity police" -F country=us -F court_level=scotus -k 5
  54. # Trials, filings, vulnerabilities
  55. deepxiv fa search trials "GLP-1 obesity" -F phase=PHASE3 -F overall_status=RECRUITING -k 5
  56. deepxiv fa search filings "CEO resignation" -F market=us -F "filed_year>=2023" -k 5
  57. deepxiv fa search cve "*" -F kev=true -F severity=critical,high --sort epss -k 10
  58. # A cited answer over one domain (agentic quota; SDK token: 3 trial asks)
  59. deepxiv fa ask "个人信息出境需要满足什么条件" -d law
  60. deepxiv fa ask "DeepSeek-R1 得到了哪些院士的正面评价" -d appraise -p paper=arxiv:2501.12948
  61. ```
  62. ### 🆕 US stocks
  63. `deepxiv fa … stocks` gives an agent point-in-time US equity data: **8,000+ tickers** with daily prices, **per-ticker news** (3.9M stories since April 2026, deduplicated and clustered into ~920k events with a tone label), SEC filings, XBRL financials, and **macro series and news** (CPI, industrial production, Treasury yields, FX, commodities, with consensus/nowcast comparisons). Every query takes `as_of`, which returns only what was visible at that moment — so backtests and "what did we know on day X" questions don't leak the future.
  64. ```bash
  65. deepxiv fa read stocks NVDA # price vs sector & SPY, recent events, news, filings
  66. deepxiv fa read stocks NVDA -e news # the ticker's news feed
  67. deepxiv fa read stocks NVDA -e financials # XBRL financials, as filed
  68. deepxiv fa read stocks NVDA -p as_of=2026-09-04 # the same view as of a past day (here: the Hugging Face deal 8-K)
  69. deepxiv fa search stocks "AI data center power" -F source=stocks # which tickers a theme touches
  70. deepxiv fa search stocks "chip export restrictions" -k 5 # news + events + filings, ranked together
  71. deepxiv fa search stocks "earnings guidance" -F ticker=TSLA -F source=events
  72. deepxiv fa read stocks digest:macro # today's macro releases, rates, FX, commodities
  73. deepxiv fa read stocks macro:CPIAUCSL # a macro series with vintages and nowcast
  74. # Or just ask: the agent pulls prices, news, events, filings, financials and macro, and cites every record
  75. deepxiv fa ask "CoreWeave 最近一周为什么大跌?结合价格、新闻和财报说明" -d stocks
  76. ```
  77. **`fa ask -d stocks` is live**: ask about a theme, a ticker's moves, events, insider trades, financials or macro, and get an answer that cites upstream record ids (`[stocks:px:CRWV:2026-10-16]`, `[stocks:news:…]`, `[stocks:fin:…]`). Questions about a past moment are answered as of that moment. Answers end with the data sources and a not-investment-advice note. `deepxiv fa spec stocks` lists sources, filters and extras.
  78. DeepXiv itself stays focused on agentic data services for academic papers, and **will remain free**. If you need higher limits on the 1stAuthor domains, see [1stauthor.com](https://1stauthor.com).
  79. > 1stAuthor is in closed beta. Questions or problems? [Open an issue](https://github.com/DeepXiv/deepxiv_sdk/issues).
  80. ## Install
  81. ```bash
  82. pip install deepxiv-sdk
  83. ```
  84. `deepxiv fa` (1stAuthor domains, including talent) needs 1.2.0 or later: `pip install -U deepxiv-sdk`.
  85. `deepxiv` auto-registers a token on first use. Agentic commands (`ask`, `fa ask`) need a registered key; the auto-registered token gets 3 lifetime trial `fa ask` calls. Get a key at [data.rag.ac.cn/register](https://data.rag.ac.cn/register), then:
  86. ```bash
  87. deepxiv config --token YOUR_REGISTERED_KEY
  88. ```
  89. During the beta, a registered key uses the 1stAuthor domains on the same terms as 1stAuthor's own users:
  90. - **`deepxiv fa ask`** draws on a **daily ask allowance** that resets at 00:00 UTC. When it runs out, you'll see "今日 AI 问答额度已用完" (today's ask allowance is used up). There is also a site-wide daily total, which can run out early at peak times.
  91. - **`deepxiv fa search` / `read`** are fair use, rate-limited to 60 requests a minute, 5 at once and 5,000 a day.
  92. - **`talent`** also has a daily read limit on profile content (search and cards aren't capped).
  93. - **The auto-registered token** stays as it is: 3 lifetime trial `fa ask` calls, then a prompt to register at [data.rag.ac.cn/register](https://data.rag.ac.cn/register).
  94. - **On a 429**, back off for the number of seconds in `Retry-After`.
  95. `deepxiv ask` (papers) is DeepXiv's own and unchanged: 300 calls/day on a registered key, on a pool separate from the general daily limit.
  96. ## Usage
  97. One investigation, start to finish. You've heard speculative decoding got much faster this year and want to know what's real.
  98. **1. Ask the literature.** Start with the question, not a keyword. The service picks its own tools, reads paper bodies, and cites what it used.
  99. ```bash
  100. deepxiv ask "what speedup does speculative decoding report on HumanEval in 2025"
  101. ```
  102. ```
  103. DEER reports a 5.54× speedup on HumanEval (with Qwen3-30B-A3B as the target
  104. model), compared to EAGLE-3's 2.41× on the same benchmark [arXiv:2512.15176].
  105. 📚 Sources (1 cited, 10 retrieved — use --all-sources for the rest):
  106. 1. [2512.15176] DEER: Draft with Diffusion, Verify with Autoregressive Models
  107. ```
  108. <p align="center">
  109. <img src="./assets/demo.gif" width="100%">
  110. <br>
  111. <em><code>deepxiv ask</code> — a question in, a cited answer streaming out</em>
  112. </p>
  113. The answer goes to stdout and sources to stderr, so `deepxiv ask "…" > answer.md` captures just the answer. Add `--effort high` when a question spans several papers.
  114. **2. Read the paper it cited — in layers.** Never load a whole paper to answer a question about one section.
  115. ```bash
  116. deepxiv paper 2512.15176 --brief # title, TLDR, keywords, citations — worth reading?
  117. deepxiv paper 2512.15176 --head # section list + where the tokens are
  118. deepxiv paper 2512.15176 --section Experiment # read only that
  119. ```
  120. Each step costs an order of magnitude more than the last, so you stop as soon as you have your answer. Take section names from `--head` — papers don't share a common outline. `--preview` gives ~10k chars; no flag at all gives the full markdown.
  121. **3. Widen it into a search.** Once you know what you're looking for, filter for the rest.
  122. ```bash
  123. deepxiv search "speculative decoding" --date-from 2025-01 --min-citations 20 --limit 10
  124. ```
  125. Filters combine with `AND` — `--authors`, `--orgs`, `--categories`, `--venue`/`--venue-year`, dates, citation floors. Stack too many and you'll legitimately get zero results; loosen one.
  126. **4. Find the people behind it.** A method is worth more when you know whose lab it comes from and what else they've built.
  127. ```bash
  128. deepxiv fa search talent "speculative decoding" --top-k 5 # by topic
  129. deepxiv fa search talent "Geoffrey Hinton" # or by name
  130. deepxiv fa search talent "deep learning" -F "org=University of Toronto" -F "h_index>=50"
  131. deepxiv fa read talent 15023 # profile: bio, education, career, top papers, metrics
  132. deepxiv fa read talent 15023 --level full # the full researched profile (markdown)
  133. ```
  134. Names (Chinese, pinyin, English) and topics both work. IDs from `search` feed `read`; `deepxiv fa spec talent` lists the filters and what each read level costs. `deepxiv talent search|survey` still work as deprecated aliases, but IDs from the old index don't carry over.
  135. The same verbs cover 1stAuthor's other domains (`deepxiv fa domains` lists them). **Full guide: [FA.md](FA.md).**
  136. ```bash
  137. deepxiv fa search law "个人信息保护" --top-k 5
  138. deepxiv fa ask "个人信息出境需要满足什么条件" --domain law # agentic quota; SDK token: 3 trial asks
  139. ```
  140. **5. Step off arXiv when the question isn't academic.** Licensing, pricing, who shipped what last week — same command, different backend.
  141. ```bash
  142. deepxiv ask "which inference providers support speculative decoding today" --web
  143. deepxiv ask "NeurIPS 2025 best paper" --web --search-type news
  144. ```
  145. The web backend reads *cached* page bodies. Pages read in full are marked 📄, snippet-only ones 🔗 — weigh them accordingly.
  146. **In Python**, the same pipeline is `Reader`:
  147. ```python
  148. from deepxiv_sdk import Reader
  149. reader = Reader(token="YOUR_REGISTERED_KEY") # Reader takes the token explicitly
  150. answer = reader.agent_search("what speedup does DEER report on HumanEval")["answer"]
  151. method = reader.section("2512.15176", "Method")
  152. people = reader.fa().search("talent", "speculative decoding", top_k=5)
  153. ```
  154. ## Documentation
  155. - **[USAGE.md](USAGE.md)** — full CLI reference, the Python API, streaming, error handling, batching, and the built-in research agent. ([中文](USAGE.zh.md))
  156. - **[FA.md](FA.md)**: `deepxiv fa`, the 1stAuthor domains (talent, law, cases, trials, drugs, grants, filings, standards, CVEs, …): commands, filters, prices, errors, Python. ([中文](FA.zh.md))
  157. - **[skills/deepxiv-cli/SKILL.md](skills/deepxiv-cli/SKILL.md)** — drop-in operating instructions for coding agents. Two worked workflows also ship as skills: [trending digest](skills/deepxiv-trending-digest/SKILL.md), [baseline table](skills/deepxiv-baseline-table/SKILL.md).
  158. - **`skills/deepxiv-fa-*`** — one skill per 1stAuthor domain (e.g. [stocks](skills/deepxiv-fa-stocks/SKILL.md), [talent](skills/deepxiv-fa-talent/SKILL.md), [law](skills/deepxiv-fa-law/SKILL.md)): filters, read levels, extras and verified recipes as `deepxiv fa` commands, generated from the live spec by `scripts/gen_fa_skills.py`. Index: [fa-domains.md](skills/deepxiv-cli/references/fa-domains.md).
  159. - **[examples/](examples/)** — runnable scripts for each entry point.
  160. Also available: PubMed Central, bioRxiv/medRxiv, trending papers, and per-paper social metrics — see [USAGE.md](USAGE.md#other-sources).
  161. ## Citation
  162. If DeepXiv is useful in your work, please cite the technical report:
  163. ```bibtex
  164. @article{qian2026deepxiv,
  165. title = {DeepXiv-SDK: An Agentic Data Interface for Scientific Literature},
  166. author = {Qian, Hongjin and Xia, Ziyi and Liu, Ze and Chen, Jianlyu and
  167. Luo, Kun and Qin, Minghao and Li, Chaofan and Xiong, Lei and
  168. Lan, Junwei and Wang, Sen and Liang, Zhengyang and Shao, Yingxia and
  169. Lian, Defu and Liu, Zheng},
  170. journal = {arXiv preprint arXiv:2603.00084},
  171. year = {2026},
  172. url = {https://arxiv.org/abs/2603.00084}
  173. }
  174. ```
  175. ## License & support
  176. MIT — see [LICENSE](LICENSE).
  177. - 🐛 **Issues**: [github.com/DeepXiv/deepxiv_sdk/issues](https://github.com/DeepXiv/deepxiv_sdk/issues)
  178. - 📧 **Higher limits, bulk use or donations**: email `tommy[at]chien.io` with your use case

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.