符合nature论文学术表达和科研绘图的Skill
27.5k
Stars
1.7k
Forks
0
Open issues
19
Contributors
AI Analysis
Nature-Skills is a collection of AI agent skills designed specifically for academic researchers conducting paper analysis, scientific writing, figure generation, and peer review workflows aligned with Nature journal standards. It serves researchers and academics globally who use AI coding assistants (Claude, Codex, OpenCode) to streamline their scientific publication pipeline; it is not intended as a general-purpose tool for non-research domains.
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
AI skill pack for Nature-style academic writing and scientific figure generation, targeting Chinese researchers using Codex agents
nature-skills is a curated collection of reusable 'skills' (structured prompt-plus-script modules) designed for OpenAI Codex and compatible AI agents. It targets academic researchers—primarily Chinese-speaking—who want to automate repetitive scientific tasks: polishing manuscripts to Nature journal style, generating publication-quality figures, formatting citations, drafting reviewer responses, and converting papers to slides. Each skill is a self-contained directory with a SKILL.md prompt template, optional scripts, and shared assets. The project is aimed at reducing friction in the academic publishing pipeline by wrapping expert conventions into callable AI instructions.
Created in late April 2026, the project emerged in the same period that OpenAI Codex agent skills became a publicly usable pattern. It grew rapidly from zero to ~21,000 stars within roughly seven weeks, driven largely by Chinese research communities and social media promotion on Douyin (TikTok China).
The creator actively promoted the project via Douyin video tutorials and a paid knowledge community (知识星球), targeting Chinese graduate students and researchers. The surge to 21,000+ stars in under two months—with 430 stars in the last seven days as of the evaluation date—reflects strong resonance within Chinese academic social networks rather than broad international organic discovery. The 'skills' format riding the Codex wave likely amplified visibility in a moment of platform-level novelty.
No independent third-party production usage is documented. Adoption evidence comes entirely from the creator's own README narrative, Douyin community engagement, and star count. The claim that Google DeepMind referenced this project is stated only by the creator in the README and is not independently verifiable. Adoption appears real but concentrated in Chinese academic hobbyist/early-adopter communities rather than verified institutional or production research pipelines.
Appears to use a directory-per-skill layout under skills/, where each unit contains a SKILL.md (prompt definition), an optional manifest.yaml for routing, static assets, and a references/ folder. A shared _shared/ directory provides common utilities. This appears to be a prompt-engineering library with supporting scripts rather than a traditional software package—no pip-installable module or API is evident from the README.
not documented in README
Last push was 2026-06-20, the same day as evaluation—indicating active, ongoing development. The project is less than two months old, so longevity cannot yet be assessed. The creator explicitly commits to continued updates regardless of monetization, and the skill index shows multiple maturity levels (Stable, Beta, Draft), suggesting iterative development is in progress.
ADOPT IF: you are a Chinese-speaking graduate student or researcher using OpenAI Codex who routinely prepares manuscripts or figures for high-impact journals and wants to reduce repetitive formatting and writing overhead. AVOID IF: you need production-grade reliability, reproducible figure pipelines for peer scrutiny, or are working in non-Chinese academic contexts where community support is absent. MONITOR IF: you are building AI-assisted research tooling and want to track whether prompt-based academic skills mature into robust, institutionally trusted workflows.
Independent dimensions
Mainstream potential
3/10
Technical importance
4/10
Adoption evidence
2/10
- The project is tightly coupled to the OpenAI Codex skills runtime; changes to Codex's API, pricing, or skills format could break all skills simultaneously with no fallback path.
- The README contains unverifiable claims (e.g., Google DeepMind borrowing from this project) that, if inaccurate, raise questions about the creator's reliability as a narrator of the project's impact.
- Scientific figure and citation generation via prompt templates carries a meaningful risk of hallucinated references or stylistically incorrect outputs that could embarrass researchers if not manually verified.
- Community and tutorial content appear concentrated on Douyin, a platform with uncertain long-term accessibility outside China and subject to content moderation; knowledge transfer risk is real.
- At under two months old, the project has no demonstrated maintenance track record; the creator's continued commitment beyond the initial viral moment is unproven.
Likely to remain a popular reference within the Chinese Codex/academic-AI community for the near term, with gradual skill additions. May plateau or fragment as the Codex skills ecosystem itself evolves or as competing collections emerge. International mainstream adoption appears unlikely without English-first documentation and broader community infrastructure.
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- Python
- License
- Apache-2.0
- Last updated
- 10h ago
- Created
- 3mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
No open issues — clean slate.
Open pull requests
No open pull requests.
Top contributors
Recent releases
No releases published yet.
Similar repos
zLanqing/codex-claude-academic-skills
This project provides three specialized Claude Code/Codex skills for Chinese...
Master-cai/Research-Paper-Writing-Skills
This repository packages research paper writing methodology and templates as...
HughYau/AcademicForge
Academic Forge is a curated platform for installing and managing AI skills for...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
Yuan1z0825/nature-skills
this
|
27.5k | +2.2k | Python | 8/10 | 10h ago |
|
zLanqing/codex-claude-academic-skills
|
1.7k | — | Python | 7/10 | 2mo ago |
|
Master-cai/Research-Paper-Writing-Skills
|
5k | — | — | 7/10 | 2w ago |
|
HughYau/AcademicForge
|
2.3k | — | Python | 7/10 | 1d ago |
 openai/skills
|
23.5k | — | Python | 3/10 | 18h ago |
 Imbad0202/academic-research-skills
|
37.2k | — | Python | 8/10 | 2d ago |
Likely the reference implementation for the Codex skills format that nature-skills builds upon. More general-purpose and officially maintained by OpenAI; nature-skills is a domain-specific extension targeting academic research conventions, especially Nature journal style.
Another high-star skills collection from the Chinese developer community. Appears to be general-purpose rather than science-specific. Both projects compete for the same audience of Chinese Codex early adopters, but with different domain focus.
Similar format and community; domain scope not clear from available metadata. Nature-skills differentiates by its explicit academic publishing focus and tutorial ecosystem.
Appears to be a curated list or aggregator of skills rather than a domain-specific pack. More of a directory than a competing skill toolkit for science.
The incumbent workflow for Nature-style academic writing and figures involves manual LaTeX editing, matplotlib/ggplot scripting, and Zotero citation management. nature-skills attempts to automate these conventions via AI prompts, trading flexibility and precision for speed—a meaningful tradeoff for early-career researchers without deep tooling experience.
