Overview
This Skill guides an AI Agent to assist users in writingNational Natural Science Foundation of Chinagrant proposals, based on the latest official 2026 regulations, covering three project categories: Young Scientists (Category C), Excellent Young Scientists (Category B), and Distinguished Young Scholars (Category A). Unlike the existing research-grants(for overseas funding agencies such as NSF/NIH/DOE), this Skill is designed specifically for themainland China NSFC system, follows the three main body module structure after the "streamlining and quality improvement" revision (project rationale / research content / research foundation), and highlights the new 2026 review mechanisms (phased review, initial screening of the first 500 characters, AI-assistance statement).
Academic Integrity Statement: It is prohibited to use AI to directly generate grant proposal content; AI may only be used to assist with literature search, structure organization, and language polishing. All scientific viewpoints, research ideas, and innovation points must be provided by the applicant personally, and factual accuracy must be verified manually.
When to Use This Skill
This Skill should be triggered when the user's request involves the following situations:
- Writing, revising, or polishing any module of an NSFC grant proposal
- Asking about application regulations, word limits, application codes, or subcategory descriptions for the Young Scientists Fund / Excellent Young Scientists Fund / Distinguished Young Scholars Fund
- Checking whether a grant proposal complies with the 2026 new "streamlining" requirements (no more than 30 pages, no separate "research plan" section, etc.)
- Assessing whether the project rationale presents innovation within the first 500 characters (to address phased review)
- Organizing 5 representative works and writing a "statement of personal contribution"
- Writing compliance documents such as ethics review, AI-assistance statement, and human genetic resources statement
Quick Reference of Core Differences Among the Three Project Categories
| Dimension | Category C (Young Scientists Fund) | Category B (Excellent Young Scientists Fund) | Category A (Distinguished Young Scholars Fund) |
|---|---|---|---|
| Applicant age | Male ≤35 / Female ≤40 | Male ≤35 / Female ≤40 | Male ≤45 / Female ≤48 |
| Funding amount | 24300,000 | ~2 million | ~4 million (renewal 8 million) |
| Funding period | 3 years | 3 years | 5 years |
| Review focus | Innovation potential | Has achieved relatively good research results | Top-tier talent who has achieved outstanding results |
| Main body page limit | No more than 30 pages | No more than 30 pages | No more than 30 pages |
| Budget | Lump-sum system, no budget preparation required | Lump-sum system | Lump-sum system |
| Proposal structure (2026) | Project rationale + research content + research foundation | Same as left | Same as left + applicant information |
For detailed regulations, see
references/NSFC_WRITING_SPEC.md。
Workflow
Phase 1 · Information Gathering
Collect the following elements through dialogue (or structured questioning in supported tools):
【申请类型】：C类（青基）/ B类（优青）/ A类（杰青）/ A类延续
【项目名称】：20字以内名词性短语
【申请代码】：必须选至二级（4位数字），可附2~3个辅助代码
【研究期限】：C/B类3年；A类5年
【关键词】：3-5个，含研究对象、核心方法、关键科学问题
【申请人信息】：职称、单位、研究方向、年龄（核对资格）
【拟解决的核心科学问题】：1-2句话
【现有研究基础】：已发论文（按相关性排序）、初步实验数据、技术平台
【是否涉及伦理 / 人遗 / 生物安全 / 临床】：用于决定附件清单
【是否使用 AI 辅助检索】：用于撰写AI使用声明
If the user's information is incomplete, follow up on the fields that most affect the Agent's decisions (application type, core scientific question, research foundation); the rest may be filled with the placeholder "please supplement on your own".
Phase 2 · Literature Search
Refer to references/LITERATURE_SEARCH.md, and select a search strategy according to the discipline:
- If
qinyan-paper-searchSkill is installed → preferentially call its unified search interface (covering Google Scholar / PubMed / ArXiv / Wanfang) - If
WebSearch/WebFetchis available → use keyword combinations to search reviews and representative papers from the past 5 years - If no external network tool is configured → prompt the user to prepare a local PDF/Markdown literature library, based on which the Agent writes the review
Disciplinary orientation suggestions：
| Division direction | Priority search databases | Notes |
|---|---|---|
| Life Sciences / Medicine | PubMed → Web of Science → Wanfang | PubMed only accepts English queries |
| Chemistry / Materials | Web of Science → SciFinder → ArXiv | |
| Information / Mathematics / Physics | ArXiv → Google Scholar → DBLP | |
| Engineering / Environment | Web of Science → Google Scholar → Wanfang/CNKI | |
| Interdisciplinary | OpenAlex → Google Scholar | Use OpenAlex to obtain the citation network |
Phase 3 · Writing the Project Rationale (6–10 pages)
Follow the "inverted pyramid" structure of references/NSFC_WRITING_SPEC.md §3.1:
- Opening 500 words(extremely critical, the focus of the 2026 phased review initial screening):
- Sentence ①: the core scientific question of this research and the original entry point
- Paragraph ②: the essential difference from existing research
- Paragraph ③: potential scientific value
- Strictly prohibitedto pile up a lengthy conventional review at the beginning
- Research background and domestic/international status: macro → meso → micro gaps, covering core literature from the past 5 years
- Scientific question and project rationale: clearly define a "why/how"-type basic scientific question (not an engineering "how to do" one)
- Differences between Category C and Categories B/A：
- Category C focuses on the applicant'sability to independently propose questions
- Categories B/A require sufficient prior work support
Phase 4 · Writing Research Content + Research Foundation
Research content(10–14 pages) — 2026 reform requirements:
- Abolish the mandatory requirement for independent sections on "research plan / technical route / annual plan"
- Research methods must benaturally integratedinto the narrative of the research content
- All sub-content must be unified alongthe core scientific questionas the axis, progressing logically rather than being listed in parallel
- Features and innovation points are recommended to form a separate paragraph, 2–4 items, each containing the four elements [type] + [current status] + [breakthrough] + [significance]
Research foundation(6–10 pages) — the golden rules for selecting and arranging 5 representative works:
| Dimension | Requirement |
|---|---|
| Timeliness | Prioritize the past 5 years; CNS main journals/high-level sub-journals may be relaxed |
| Journal level | At least 3 papers in Zone 2 or above; avoid warning/On-hold journals |
| Author order | Category C emphasizes first authorship (≥2 unique first-author papers in Q2 or above) |
| Relevance | 5 papers revolve around the same direction, forming a "mechanism discovery → validation → expansion" chain |
| Ideal configuration | 3 directly related + 1 expanded + 1 high-impact corroborating |
Format for the personal contribution statement for representative works(2026 enhanced requirement):
"In this research, I was responsible for: [specific contribution], accounting for approximately [X]% of the workload."
Phase 5 · Compliance and Self-Check
After completing the draft, check item by item in references/QUALITY_CHECKLIST.md :
- Formal completeness: Application code, page count, attachments (ethics/human genetic resources/biosafety)
- Content quality: Opening 500 characters, whether innovation points are supported by literature comparison, relevance of representative works
- Integrity and compliance: AI-assisted declaration, over-quota check, distinction from ongoing projects
Output the final application draft (Markdown format), with a summary of the self-check results attached.
Tool compatibility notes
This Skill uses pure Markdown + YAML frontmatter,does not depend on any tool-specific syntax. Usage in each Agent:
| Tool | Loading method | Invocation suggestion |
|---|---|---|
| Claude Code | ~/.claude/skills/nsfc-proposal/, automatically recognizes frontmatter |
Triggered when the user says "help me write an NSFC Young Scientists Fund" |
| Cursor | .cursor/skills/nsfc-proposal/ or .cursorrules reference SKILL.md in |
@SKILL.md in chat |
| Codex CLI | ~/.codex/skills/nsfc-proposal/, as prompt context |
Use codex --skill nsfc-proposal or include in the initial prompt |
| OpenClaw | ~/.openclaw/skills/nsfc-proposal/ |
Same as Claude Code |
| Gemini CLI | ~/.gemini/skills/nsfc-proposal/ |
Reference the SKILL.md path in the system prompt |
Installation command:
curl -fsSL https://raw.githubusercontent.com/LeonChaoX/qinyan-academic-skills/main/install.sh | \
bash -s -- -s nsfc-proposal --tool <claude|cursor|codex|gemini|openclaw>
Referenced files
| File | Purpose |
|---|---|
references/NSFC_WRITING_SPEC.md |
Complete writing specifications (basic information, abstract, three main body modules, personal resume, special circumstances) |
references/LITERATURE_SEARCH.md |
Literature search strategies and recommended data sources for each academic division |
references/QUALITY_CHECKLIST.md |
Pre-submission self-check list covering form/content/integrity |
High-frequency error warnings
| Error | Correction |
|---|---|
| Piling up clichés ("of great theoretical significance") | Use specific data, mechanisms, and comparisons to explain significance |
| Scattered research directions (≥5 independent sub-directions) | Focus on 2-3 core questions and dig deep |
| Overly ambitious goals ("establish a complete theoretical system") | Change to specific scientific goals achievable within 3 years |
| Missing methods | Integrate key method descriptions into the research content |
| Innovation points not supported by literature comparison | Provide a "current status + breakthrough" comparative explanation for each |
| Length out of control (exceeding 30 pages) | Delete clichés; every sentence must contain substantive information |
| Representative works unrelated to the project | Select only strongly related achievements and explain personal contributions in detail |
Final reminder: The application must be written by the applicant themselves. This Skill serves only as a structural framework and writing guidance; all scientific content must come from the applicant's own research work and academic thinking. A high-quality application reflects the applicant's academic thinking and expression ability, rather than the application of a template.