A writing paradigm distilled from 16 successful NSFC applications, implementing an automated workflow of "topic analysis → policy research → data research → writing the justification".
Skill Scope
This skillonly handles the justification section of an NSFC application, 1.1 Research Significance section, including:
- Research background (including national strategy, real-world dilemmas, academic problems, this project's approach)
- Theoretical significance
- Practical significance
Trigger Conditions
Activate this skill when the user asks to write an NSFC justification, research significance, or project background, or provides a research topic and requests generation of a justification.
Execution Workflow
After receiving the materials provided by the user, strictly follow these 7 steps:
The userat minimumwill provide the following inputs: research content, research methods, technical route, features and innovations, research objectives, key scientific questions. If the input is insufficient,you must first ask follow-up questions to fill in the gapsbefore continuing.
Step 1: Read the knowledge base
Read the following files to learn the writing paradigm:
assets/templates/writing_style_guide.md—— style specification and narrative logicassets/templates/structure_template.md—— paragraph arrangement and sentence logical roles
Step 2: Topic analysis
- Parse the core keywords of the research topic
- Identify the discipline field it belongs to (Management Science and Engineering / Medicine / Economics / Information Science, etc.)
- Determine the nature of the scientific question (encouraged exploration / frontier-focused / demand-driven / interdisciplinary integration)
Step 3: Input-background anchoring
Mapping principle: let the agent know what each paragraph of the background should "point to", ensuring logical coherence. Anchoring is a "shadow correspondence" at the logical level, not "copy-paste" at the textual level — the dilemmas/deficiencies/approaches in the background should be written in the background's own language and argumentative perspective, but logically must be traceable to the subsequent inputs.
Input→Paragraph Mapping Table：
| User Input | Corresponding Background Paragraph | Writing Perspective Transformation |
|---|---|---|
| Research content | Number of dilemma items in Paragraph C | "Real-world problem" perspective, not "research content" perspective |
| Details of research methods | Number of deficiency items in Paragraph D | "Limitations of existing methods" perspective, not "this study's methods" perspective |
| Technical route | Expansion order of the approach in Paragraph E | Condense into "what to do" without saying "how to do it", using "this project intends to……" |
| Features and innovations | Items of theoretical significance | Innovation points state "what was done", theoretical significance states "what academic value it has" |
| Research objectives | Items of practical significance | Research objectives state "what is to be achieved", practical significance states "what value it has for society/industry" |
| Key scientific questions | Enumeration of scientific questions in Paragraph F | Make slight rewrites, but do not copy directly, and keep the wording reasonably close to the subsequent inputs |
Independent constraints: Paragraphs A and B (policy background + data background) are not subject to subsequent input constraints and are determined by online retrieval results.
Step 4: Online retrieval
Execute online search according to the following strategy (the agent selects available tools on its own):
Policy retrieval (see details in references/policy_search_strategy.md）
- Preferred sources: State Council, Ministry of Science and Technology, NSFC, National Bureau of Statistics, Ministry of Education, National Health Commission, Ministry of Industry and Information Technology, National Development and Reform Commission, and the competent authorities corresponding to the research topic
- Goal: Find 2-3 national/department-level policy documents directly related to the topic
- Record: policy name, year of publication, core requirements/statements
Data retrieval (see details in references/data_search_strategy.md）
- Preferred sources: National Bureau of Statistics, industry blue books, authoritative white papers, reports of international organizations, high-level review articles
- Goal: Find 3-5 statistical data points supporting the severity of the problem
- Record: year + specific value + source + original link
- ProhibitedUse of marketing articles, unsourced data, or reposted data
Step 5: Generate the project rationale outline
According to structure_template.md paragraph structure, generate the outline:
- Annotate each paragraph: what to write, what data/policy to use, the logical role of the sentence
- Ensure a one-to-one correspondence between problems and solutions
- Show the outline to the user and proceed to writing after confirmation
Step 6: Write the complete project rationale
Strictly follow writing_style_guide.md style specifications:
- Write in the order of paragraphs A→B→C→D→E→F → theoretical significance → practical significance
- Ensure each paragraph is supported by policy/data
- Explain theoretical significance point by point using the "verb + object + value" structure
- Explain practical significance progressively from the macro level to the micro level
- Avoid absolute expressions such as "first time" and "fill the gap"
- Methodological terms should only serve as background/verification means, not carry the main line of the paragraph
Step 7: Self-check and output
- According to
references/dod_checklist.mdcheck structural completeness - According to
references/boastful_expression_guidelines.mdidentify exaggerated expressions - According to
references/dimension_coverage_design.mdcheck coverage of the four dimensions - Output the final text, and also attach:
- List of policies used(policy name + year + core statement)
- List of statistical data used(data item + year + value + source + link)
Reference file index
| File | Purpose |
|---|---|
assets/templates/writing_style_guide.md |
Writing style guide (narrative logic, paragraph arrangement, sentence logical roles, style specifications) |
assets/templates/structure_template.md |
Project rationale structure template (paragraph-sentence logical skeleton) |
references/info_form.md |
Minimum information table (ask first if required fields are missing) |
references/scientific_question_guidelines.md |
Key points for writing scientific questions |
references/scientific_hypothesis_guidelines.md |
Key points for writing scientific hypotheses |
references/boastful_expression_guidelines.md |
List of prohibited exaggerated expressions |
references/dimension_coverage_design.md |
Dimension coverage check design |
references/theoretical_innovation_guidelines.md |
Guide for theoretical innovation orientation |
references/dod_checklist.md |
Definition of Done checklist |
references/methodology_term_examples.md |
Comparative examples of misuse of methodological terms |
references/policy_search_strategy.md |
Policy retrieval strategy |
references/data_search_strategy.md |
Data retrieval strategy |