**deepxiv-baseline-table**

deepxiv-sdk

**DeepXiv is an agent-first paper search and progressive reading tool.**

Install it with `pip`, start using it immediately, and let the CLI auto-register an API token on first use. No extra setup is required before your first query.

• 📚 API Documentation: https://data.rag.ac.cn/api/docs
• 🎥 Demo Video:
• 📄 Technical Report:
• 📖 中文文档: README.zh.md

---

🚀 **Live Demo**: I used vibe coding, based on deepxiv CLI, to build a [DeepResearch demo](https://demo.rag.ac.cn/) in 1 hour — feel free to try it out! A full-stack research platform is on the way. Stay tuned.

---

What DeepXiv Does

DeepXiv is built around two core workflows that matter for agents:

1. Search + Progressive Content Access
2. Trending + Popularity signals

Instead of blindly loading full papers, DeepXiv lets agents read in layers, based on token budget and task value.

Quick Start

pip install deepxiv-sdk

On first use, deepxiv automatically registers a free anonymous token (1,000 requests/day) and saves it to `~/.env`:

deepxiv search "agentic memory" --limit 5

If you want the full stack including MCP and the built-in research agent:

pip install "deepxiv-sdk[all]"

CLI-First Workflow

The CLI is the primary interface. DeepXiv is designed so agents can work like researchers: search first, judge quickly, then read only the most valuable parts.

deepxiv search "agentic memory" --limit 5
deepxiv paper 2603.21489 --brief
deepxiv paper 2603.21489 --head
deepxiv paper 2603.21489 --section Analysis

Three commands matter most for progressive reading:

• --brief: decide whether a paper is worth deeper reading
• --head: inspect structure, sections, and token distribution
• --section: read only the most valuable parts such as Introduction, Method, or Experiments

...