nature-literature-pipeline

1# Nature Literature Pipeline
2 
3A complete, production-tested automated literature pipeline. Not just "search for papers" — it's a structured engine that scores, classifies, reads, delivers, and archives research papers daily.
4 
5## What It Does
6 
7```
8Cron (daily trigger, e.g. 08:30)
9  │
10  ├─ ① SEARCH (30 candidates)
11  │   arXiv / OpenAlex / Crossref / Semantic Scholar (auto-degradation)
12  │
13  ├─ ② COARSE FILTER (30 → 5)
14  │   Six-dimension scoring: topic match × 35 + methodology × 20
15  │   + journal quality × 15 + network relevance × 10
16  │   + applied value × 10 + archival value × 10
17  │
18  ├─ ③ FINE READ (top 5)
19  │   Abstract-level or full-text. Source level tagged:
20  │   Full-text / Abstract only / Metadata only
21  │
22  ├─ ④ DELIVER
23  │   Formatted digest to Feishu/Telegram/etc.
24  │   🏅 rank | title | journal | ⭐ score | 💡 one-liner
25  │   🔬 methods | 📊 key results | 🧭 commentary
26  │
27  └─ ⑤ ARCHIVE
28      DOI/arXiv de-dup → classify → write notes → update index
29```
30 
31## Quick Start
32 
33After installing, tell your agent:
34 
35```
36My research area is [X], keywords: [Y], deliver to [feishu group name], archive to [path]
37```
38 
39The agent will configure keywords, delivery target, and archive path automatically.
40 
41Then set up a daily cron job:
42 
43```
44Set up a daily literature push at 08:30 Beijing time, 30 candidates, top 5 delivered
45```
46 
47## Architecture
48 
49The skill is organized in two layers:
50 
51| Layer | Purpose | Files |
52|-------|---------|-------|
53…