nsfc-benzi-audit

Name: nsfc-benzi-audit
Author: jiankang1991

28stars3forks

How to install nsfc-benzi-audit?

jiankang1991/nsfc-benzi-audit/nsfc-benzi-audit

$npx -y skills add jiankang1991/nsfc-benzi-audit --skill nsfc-benzi-audit

Installs into the current project.

Summary

Use when a user wants applicant-facing diagnosis and revision advice on a Chinese NSFC (国自然) application draft — asks for 本子把脉, 本子体检, 国自然申请书修改建议, NSFC benzi audit, 帮我看国自然本子, 标书逻辑诊断, 青年/面上/地区基金申请书修改, 对照已中本子, or 从中标样本提炼写法规律 — or wants critique of the title, abstract, key scientific

Files · 1

View on GitHub

SKILL.md

1	# NSFC Benzi Audit
2
3	Use this skill to produce applicant-facing diagnosis and revision advice for NSFC application drafts. The goal is to expose logic breaks, weak scientific-question framing, mismatched sections, and high-impact fixes before submission.
4
5	Do not write a formal peer-review opinion unless the user explicitly asks for communication review; route that to `nsfc-review`. Do not fabricate facts, papers, project histories, budgets, or official rule details. Keep final advice grounded in the draft text and clearly mark uncertain extraction/OCR issues.
6
7	## Workflow
8
9	1. Locate and extract the draft.
10	- For PDF input, use PDF extraction/OCR as needed. Prefer existing extracted Markdown such as `full.md` or `output.md` when available.
11	- For DOCX input, extract text while preserving headings and tables where possible.
12	- For extracted Markdown/text folders, prefer `output.md`, `full.md`, or the largest readable Markdown/text file; inspect images only when visual logic diagrams or tables matter.
13	- If the file is scanned or extraction is noisy, state the limitation in the report and avoid treating OCR artifacts as applicant mistakes.
14
15	2. Identify the review scope before judging.
16	- Extract project category, research attribute, application code, title, abstract, keywords, applicant/team context, and section boundaries.
17	- If the user asks for a quick pass, inspect title, abstract, scientific questions, research contents, innovations, and research basis first.
18	- If the user asks for full diagnosis, inspect the whole application by section.
19
20	3. Load the right references.
21	- Always read `references/benzi-logic.md` before diagnosing logic or writing suggestions.
22	- Read `references/audit-surfaces.md` for full diagnosis, structure/form checks, figure/readability checks, literature-current-status checks, or policy-risk triage.
23	- Read `references/exemplar-learning.md` when the user provides already-funded/successful examples, asks to compare with "中的本子"/"中标本子", or asks to improve this skill from sample applications.
24	- Read `references/information-communication.md` when the draft or provided examples involve information science, communication networks, optical networks, computer networks, data centers, remote sensing information processing, applied AI, network security, quantum communication, or related information-engineering directions.
25	- Read `references/geospatial-remote-sensing.md` when the draft or provided examples involve remote sensing, GIS, geospatial intelligence, DEM/terrain/geomorphology, spatial databases, point clouds, SAR/optical/hyperspectral imagery, video GIS, camera networks, POI/trajectory/location data, city 3D modeling, or geospatial knowledge graphs.
26	- Read `references/medical-biomedical.md` when the draft or provided examples involve medicine, clinical research, biomedicine, disease mechanisms, patient cohorts, specimens, animal/cell/organoid models, biomarkers, diagnostics, therapy/intervention, immunology, ethics, or biosafety.
27	- Read `references/current-rules.md` when checking current-year compliance, research attributes, application-code risk, budget/ethics/scientific-integrity issues, or anything tied to official NSFC rules.
28	- Use `assets/report-template.md` as the output shape unless the user requests another format.
29
30	4. If successful examples are provided, separate exemplar learning from target diagnosis.
31	- Treat funded examples as pattern evidence, not as text to copy or proof of causality.
32	- Anonymize names, project numbers, institutions, unpublished data, and sensitive achievements before extracting patterns.
33	- Prefer patterns repeated across matched examples: same project type, similar discipline/application code, similar research attribute, or comparable career stage.
34	- Apply exemplar patterns as contrastive questions: what does the target draft fail to make visible that successful examples make visible?
35
36	5. Build the one-page logic map.
37	- Extract the draft's core logic elements: object/scenario, focused problem or goal, method/path, distinctive feature or innovation, and data/validation loop.
38	- Extract the abstract logic chain: object/problem, method/goal, contents/innovation, achievement/significance.
39	- Map these terms across title, abstract, rationale, research contents, scientific questions, innovation, feasibility,

People who liked this also viewed

6 picks

Type

nsfc-benzi-audit

28stars3forks

View on GitHub

How to install nsfc-benzi-audit?

jiankang1991/nsfc-benzi-audit/nsfc-benzi-audit

$npx -y skills add jiankang1991/nsfc-benzi-audit --skill nsfc-benzi-audit

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# NSFC Benzi Audit
2
3	Use this skill to produce applicant-facing diagnosis and revision advice for NSFC application drafts. The goal is to expose logic breaks, weak scientific-question framing, mismatched sections, and high-impact fixes before submission.
4
5	Do not write a formal peer-review opinion unless the user explicitly asks for communication review; route that to `nsfc-review`. Do not fabricate facts, papers, project histories, budgets, or official rule details. Keep final advice grounded in the draft text and clearly mark uncertain extraction/OCR issues.
6
7	## Workflow
8
9	1. Locate and extract the draft.
10	- For PDF input, use PDF extraction/OCR as needed. Prefer existing extracted Markdown such as `full.md` or `output.md` when available.
11	- For DOCX input, extract text while preserving headings and tables where possible.
12	- For extracted Markdown/text folders, prefer `output.md`, `full.md`, or the largest readable Markdown/text file; inspect images only when visual logic diagrams or tables matter.
13	- If the file is scanned or extraction is noisy, state the limitation in the report and avoid treating OCR artifacts as applicant mistakes.
14
15	2. Identify the review scope before judging.
16	- Extract project category, research attribute, application code, title, abstract, keywords, applicant/team context, and section boundaries.
17	- If the user asks for a quick pass, inspect title, abstract, scientific questions, research contents, innovations, and research basis first.
18	- If the user asks for full diagnosis, inspect the whole application by section.
19
20	3. Load the right references.
21	- Always read `references/benzi-logic.md` before diagnosing logic or writing suggestions.
22	- Read `references/audit-surfaces.md` for full diagnosis, structure/form checks, figure/readability checks, literature-current-status checks, or policy-risk triage.
23	- Read `references/exemplar-learning.md` when the user provides already-funded/successful examples, asks to compare with "中的本子"/"中标本子", or asks to improve this skill from sample applications.
24	- Read `references/information-communication.md` when the draft or provided examples involve information science, communication networks, optical networks, computer networks, data centers, remote sensing information processing, applied AI, network security, quantum communication, or related information-engineering directions.
25	- Read `references/geospatial-remote-sensing.md` when the draft or provided examples involve remote sensing, GIS, geospatial intelligence, DEM/terrain/geomorphology, spatial databases, point clouds, SAR/optical/hyperspectral imagery, video GIS, camera networks, POI/trajectory/location data, city 3D modeling, or geospatial knowledge graphs.
26	- Read `references/medical-biomedical.md` when the draft or provided examples involve medicine, clinical research, biomedicine, disease mechanisms, patient cohorts, specimens, animal/cell/organoid models, biomarkers, diagnostics, therapy/intervention, immunology, ethics, or biosafety.
27	- Read `references/current-rules.md` when checking current-year compliance, research attributes, application-code risk, budget/ethics/scientific-integrity issues, or anything tied to official NSFC rules.
28	- Use `assets/report-template.md` as the output shape unless the user requests another format.
29
30	4. If successful examples are provided, separate exemplar learning from target diagnosis.
31	- Treat funded examples as pattern evidence, not as text to copy or proof of causality.
32	- Anonymize names, project numbers, institutions, unpublished data, and sensitive achievements before extracting patterns.
33	- Prefer patterns repeated across matched examples: same project type, similar discipline/application code, similar research attribute, or comparable career stage.
34	- Apply exemplar patterns as contrastive questions: what does the target draft fail to make visible that successful examples make visible?
35
36	5. Build the one-page logic map.
37	- Extract the draft's core logic elements: object/scenario, focused problem or goal, method/path, distinctive feature or innovation, and data/validation loop.
38	- Extract the abstract logic chain: object/problem, method/goal, contents/innovation, achievement/significance.
39	- Map these terms across title, abstract, rationale, research contents, scientific questions, innovation, feasibility,

People who liked this also viewed

6 picks

Type