nature-literature-pipeline (nature-skills): a Claude skill that finds, scores and delivers new papers every day
nature-literature-pipeline makes Claude search for new papers on your topic each day, score them, summarize the best five and send a digest.
- Installs
- 12.2K
- on skills.sh
- Stars
- 51.7K
- on GitHub
- Files
- 11
- besides SKILL.md
- Updated
- Nov 2026
- last repo commit
Tell Claude your research area, keywords, delivery target and archive path, then ask for a daily push. Each run searches arXiv, OpenAlex, Crossref and Semantic Scholar for 30 candidates and scores them out of 100 on six weighted dimensions. It reads the top 5 at abstract or full-text level, sends a formatted digest, and files notes with duplicate checks.
Delivery goes to Feishu, Telegram or another messaging tool you set up. The daily run depends on a local cron job, so the machine has to be on. The skill bundles no scripts. SKILL.md says the pipeline writes to a raw/ literature folder only, and leaves your wiki alone without your approval.
Good for
- Researchers who want a daily or weekly feed of new papers in one topic
- Lab groups who share a scored digest in Feishu or Telegram
- Anyone building a deduplicated literature archive with notes
Install nature-literature-pipeline
Claude Code
- 1
Download the skill folder from https://github.com/yuan1z0825/nature-skills/tree/HEAD/skills/nature-literature-pipeline.
- 2
Put it at ~/.claude/skills/nature-literature-pipeline/ so that SKILL.md sits directly inside.
- 3
Start a new session; the skill loads when its description matches the task.
The Claude app ZIP upload is not offered here, because the daily run needs a local scheduler and a messaging setup. The related skills named in SKILL.md come from other sources and are not installed.
Built from the package data we hold. Not yet tested by us.
What's in the skill
When it loads
Complete automated literature discovery pipeline: multi-source search → six-dimension scoring → fine reading → formatted delivery → archival. Combines a configurable engine with daily cron-driven application layer. Works with Feishu, Telegram, or any messaging platform.
Sections in SKILL.md
- What It Does
- Quick Start
- Architecture
- Configuration
- Built-in Safeguards
- Related Skills
- References
- Pitfalls
Key facts
- Installs on skills.sh
- 12,205 as of 2026-11-02
- GitHub stars
- 51,696 (whole repo) as of 2026-11-02
- Last repo commit
- 2026-11-01
- Folder contents
- 12 files including SKILL.md: 6 reference files, a push template, 2 READMEs, a manifest and agents/openai.yaml
- skills.sh audits
- Gen Agent Trust Hub pass, Socket pass, Snyk warn
What it includes
Pulls 30 candidates from arXiv, OpenAlex, Crossref and Semantic Scholar.
Network ·
SearchScores six weighted dimensions and keeps the top 5.
Coarse filterReads the top papers at full-text or abstract level and tags the depth.
Network ·
Fine readSends a ranked digest with scores, methods, results and comments to chat.
Network ·
DeliverDe-duplicates by DOI, arXiv ID and OpenAlex ID, writes notes and updates an index.
Write ·
ArchiveCreates, verifies and recovers the scheduled job.
Runs code ·
Cron setupMethod for finding research gaps through a systematic survey.
Gap analysisWorkflow for compiling a literature review from notes and Zotero.
Files ·
Review compilationKeywords, scoring weights, classification tiers, delivery target and archive path.
Configuration
From the official docs
Permissions
Medium risk
The skill sets up a recurring job that searches the web, posts digests to a chat and writes notes to a local folder. Check the delivery target and archive path before you turn it on.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
It creates a recurring job that posts to a chat
The daily cron job delivers a digest to the Feishu, Telegram or other target you name. Test with a private chat first, and avoid putting sensitive project details in your keywords.
It writes notes to a local folder
SKILL.md says the pipeline writes to a raw/ literature folder only, and that wiki changes are manual. Point the archive path at a folder you want filled.
Your keywords go to outside search services
The search stage queries arXiv, OpenAlex, Crossref and Semantic Scholar with your keywords.
Technical details
| Talks to | arXiv, OpenAlex, Crossref and Semantic Scholar. your messaging platform (Feishu, Telegram or similar) |
|---|
Questions
What does nature-literature-pipeline do?
It runs a repeatable literature routine: search several sources, score candidates, read the best ones, send a formatted digest and archive notes. The README separates it from nature-academic-search, which handles one-off searches.
How do I install it?
Use the install box above. It is one folder in the nature-skills repo. After installing, tell your agent your research area, keywords, delivery target and archive path, and then ask for a daily push.
How does the scoring work?
Each paper gets a score out of 100. Topic match counts 35, methodology 20, journal quality 15, network relevance 10, applied value 10 and archival value 10. Each dimension is capped, the total is recalculated, and a topic score under 10 rejects the paper.
Where does the digest go?
To Feishu, Telegram or another messaging tool you configure. SKILL.md says it works with any messaging platform, and the README says delivery depends on your local cron, a messaging interface or your own scheduler.
Does it need an API key?
No key is mentioned. Messaging delivery and the scheduler need whatever setup those tools require. The README says automatic delivery depends on a local cron, a messaging interface or your own scheduler.
Is it safe?
It sets up a recurring job and posts digests to a chat you choose, and it writes notes into a local archive folder. SKILL.md says it writes to a raw/ folder only and does not change your wiki without approval. Check the delivery target before you turn it on.
What happens when full text is not available?
The digest tags each paper as Full-text, Abstract only or Metadata only, so you can see how deeply it was read. The README says high scores still need human reading.
Which other skills does it mention?
It names nature-academic-search, nature-citation, zotero and arxiv as related. They are optional companions from other sources, and the install box does not add them.
Information
updated Nov 2, 2026
- Repository
- yuan1z0825/nature-skills
- Folder
skills/nature-literature-pipeline
Sources
Docs, issue trackers and discussions this page is built from.
Show all 6Hide
- nature-skills README · github.com
- nature-literature-pipeline SKILL.md · github.com
- cron-setup.md · github.com
- scoring-system.md · github.com
- nature-literature-pipeline README · github.com
- nature-literature-pipeline on skills.sh · skills.sh