What does this skill do, and when should you use it?
The Nature Literature Pipeline is a complete, production-tested automated literature pipeline. It's not just a paper search tool—it's a structured engine that scores, classifies, reads, delivers, and archives research papers every day. The skill integrates multiple academic search sources (arXiv, OpenAlex, Crossref, Semantic Scholar), filters candidates using a six-dimension scoring system, provides abstract-level or full-text fine reading, and pushes formatted digests to messaging platforms like Feishu or Telegram while handling de-duplication and archival. It follows an engine/application layer architecture, allows flexible configuration of keywords, scoring weights, and delivery targets, and comes with built-in safeguards such as score validation, triple de-duplication, graceful degradation, and read-only archiving.
The skill implements a daily literature pipeline with these concrete operations:
- Downloads 30 candidate papers from arXiv, OpenAlex, Crossref, and Semantic Scholar via cron trigger (e.g., daily at 08:30), with automatic degradation if a source is unavailable.
- Scores candidates using six dimensions (topic match 35, methodology 20, journal quality 15, network relevance 10, applied value 10, archival value 10) and selects top 5.
- Performs abstract-level or full-text fine reading on high-scoring papers, tagging source level (Full-text/Abstract/Metadata).
- Formats a daily digest with rank, title, journal, score, one-liner, methods, key results, and commentary, then sends it to Feishu/Telegram/etc.
- Deduplicates using DOI/arXiv IDs, classifies, writes notes, and updates an index in a local vault/wiki directory.
- A researcher receives a daily feed of relevant papers in their field directly on Feishu, staying up-to-date without manual searching.
- A lab group sets up a daily cron job to monitor high-relevance papers in their domain and avoid missing important breakthroughs.
- A literature hoarder automatically archives pipeline outputs into Obsidian or a local vault, building a personal knowledge base.
- A researcher preparing a literature review uses the pipeline's scoring and classification to quickly filter a batch of high-value papers.
- A researcher who prefers Telegram sets the delivery target to a Telegram channel for on-the-go access to new literature.
How do you install this skill?
- License inconsistency: SKILL.md declares MIT while repository is Apache-2.0, needs clarification.
- Dependence on overseas academic services (arXiv, OpenAlex etc.) may be restricted in mainland China, affecting usability.
- Lack of user confirmation and rollback mechanisms for external side effects, risking unintended automatic pushes.
- Documentation lacks version history and changelog; maintenance responsibility unclear.
- Shell / CLI
- Network access
- Local filesystem
Node.js 18+
This skill lives in the nature-skills repository under skills/nature-literature-pipeline/, part of a collection of 19 skills.
- Clone the repository:
git clone https://github.com/Yuan1z0825/nature-skills.git - Navigate to the skill folder:
cd skills/nature-literature-pipeline - Copy the entire skill directory to your agent's skill folder (e.g.,
~/.claude/skills/or~/.codex/skills/), ensuring you preserve the complete structure includingSKILL.md,references/,templates/, etc.
Alternatively, use the npx skills add command: npx skills add Yuan1z0825/nature-skills --agent codex --skill nature-literature-pipeline --yes --copy.
Note: After installation, you'll need to configure cron and a messaging target (e.g., Feishu). Refer to references/cron-setup.md for details.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/Yuan1z0825/nature-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/nature-literature-pipeline" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- My research area is [X], keywords: [Y], deliver to [feishu group name], archive to [path]
- Set up a daily literature push at 08:30 Beijing time, 30 candidates, top 5 delivered
After installation, tell the agent your research area, keywords, and delivery target, for example:
"My research area is [X], keywords: [Y], deliver to [feishu group name], archive to [path]" .
Then set up a daily cron job:
"Set up a daily literature push at 08:30 Beijing time, 30 candidates, top 5 delivered" .
The agent will automatically run the pipeline following the SKILL.md workflow. You can also trigger a manual run to test it, or refer to references/review-compilation-workflow.md for a concentrated literature review.
What are this skill's strengths and limitations?
- Production-tested complete pipeline covering search, scoring, fine reading, delivery, and archival.
- Multi-source search (arXiv, OpenAlex, Crossref, Semantic Scholar) with graceful degradation for reliability.
- Configurable six-dimension scoring system adaptable to different fields.
- Triple de-duplication and read-only archiving protect data integrity and knowledge base safety.
- Supports Feishu, Telegram, and other messaging platforms for flexible integration into daily workflows.
- Depends on cron scheduling; if the machine is off, the pipeline doesn't run.
- Configuration requires some technical knowledge, especially cron and messaging platform integration.
- Scoring and classification rules need periodic adjustment to avoid keyword drift or score inflation.
- No automated test suite, which may lead to unexpected issues in new environments or edge cases.
- Archival only writes to the raw/ directory; wiki integration is manual.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Nature Literature Pipeline this page | 49 · Use with care | ★ 47k | 3d ago | Apache-2.0 |
| Academic Search & Citation Audit | 58 · Recommended | ★ 47k | 3d ago | Apache-2.0 |
| Paper Search | 51 · Use with care | ★ 61 | 6mo ago | MIT |
| Exa Deep Research Orchestrator | 57 · Use with care | ★ 5.1k | 4d ago | MIT |
| Exa Agent Research Skill | 56 · Use with care | ★ 5.1k | 4d ago | MIT |
It complements the nature-academic-search skill in the same repository: nature-academic-search is for ad-hoc literature searches, while this skill adds structured daily automation. Compared to generic RSS or Google Scholar alerts, this skill is more structured and automated, but its applicability is limited to research contexts.
How did FollowSkills review this skill?
The skill explicitly states it writes only to raw/ directory and does not modify wiki/knowledge base, indicating least-privilege and data-flow transparency. However, there is no explicit user confirmation step (e.g., before pushing), no isolation or rollback for external side effects (e.g., scheduled message pushes), and no mention of secure handling of sensitive data (e.g., API credentials). Therefore not full marks, partial deduction.
Static review cannot verify key path reproducibility, and there are no test suites for this skill. Documentation provides detailed degradation strategies and failure recovery steps, but lacks automated tests and clear examples of feedback on abnormal input, so reliability score is low.
The skill has clear objectives, applicable scenarios and boundaries (e.g., Chinese support, marking when full text unavailable), but lacks explicit non-fit scenarios and trigger conditions, which may cause misuse. It depends on multiple overseas academic services but does not address network reachability for mainland-China users, limiting environment fit.
Documentation is well-structured with Chinese/English READMEs and layered architecture, but lacks explicit version history and changelog. Ownership is listed in README but no clear maintenance responsibility and update path; License inconsistency exists (SKILL.md says MIT, repo Apache-2.0).
Value proposition is clear, but static review cannot verify actual output quality. Documentation provides detailed output templates but no real-world examples or delivery instances, so cannot confirm direct usability, hence limited score.
Lack of automated tests or third-party execution evidence for this skill; documentation is mostly descriptive, and evidence of verification is insufficient, so score is low.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →