Author Scientific Manuscripts with Visuals

Scientific Writing is the core skill for a deep research and writing tool, combining AI-driven deep research with well-formatted written output - every document is backed by literature search and verified citations through a companion research-lookup skill. It enforces one critical principle throughout: always write in full paragraphs with flowing prose, never bullet points, using a two-stage process of first building section outlines with key points via research-lookup, then converting those outlines into complete paragraphs.

Manuscript structure follows IMRAD (Introduction, Methods, Results, And Discussion) for standard research papers, with the Introduction establishing context and gaps, Methods detailing design/population/procedures/analysis for reproducibility, Results presenting findings objectively without interpretation, and Discussion interpreting results, acknowledging limitations, and proposing future directions - alongside support for alternative formats (narrative/systematic/scoping reviews, case reports, meta-analyses, methods papers/protocols).

Section-specific guidance covers abstract composition (100-250 words, structured or unstructured), introduction development (establishing importance, systematic literature review, identifying gaps, stating hypotheses), methods documentation (reproducibility through detailed sample description, statistical justification, ethical approval statements), results presentation (logical flow, statistical significance with effect sizes, figure/table integration), and discussion construction (relating results to questions, comparing with existing literature, honest limitations, practical implications).

Citation management covers five major styles - AMA (numbered superscript, medicine), Vancouver (numbered brackets, biomedical), APA (author-date, social sciences), Chicago (notes-bibliography or author-date), and IEEE (numbered brackets, engineering/CS) - with best practices to cite primary sources, favor recent literature (5-10 years for active fields), balance citation distribution, and verify every citation against its original source.

A mandatory visual requirement runs throughout: every scientific paper must include a graphical abstract (a landscape 1200x600px visual summary of the workflow and key findings, generated first via a generate_schematic.py script) plus additional AI-generated figures via companion scientific-schematics and generate-image skills. Minimum figure counts are specified by document type - 5 for research papers (6-8 recommended), 4 for literature reviews, 20 for market research, 1 per slide for presentations, 6 for posters, 4 for grants, 3 for clinical reports - covering study design flowcharts (CONSORT/PRISMA/STROBE), conceptual diagrams, experimental workflows, data pipelines, and photorealistic or illustrative images, with generated figures automatically checked for accessibility (colorblind-friendly, high contrast).

It also applies study-specific reporting guidelines - CONSORT for randomized trials, STROBE for observational studies, PRISMA for systematic reviews - and covers figures-vs-tables decisions (tables for precise numerical/complex datasets, figures for trends and relationships).

Use this skill when writing or revising any section of a scientific manuscript, structuring a paper with IMRAD or another standard format, formatting citations in a specific style, creating or improving figures and tables, applying a study-specific reporting guideline, drafting an abstract for journal submission, or addressing reviewer comments.

Do not use it to submit a manuscript with bullet-point prose or without the mandatory graphical abstract and minimum figure count for the document type - both are treated as non-optional requirements, not stylistic suggestions.

Input: research content, data, and findings to be structured into a manuscript, plus the target citation style and reporting guideline.

Output: a full-prose manuscript with generated figures, starting with the graphical abstract:

python scripts/generate_schematic.py "Graphical abstract for [paper title]: [brief description showing workflow from input -> methods -> key findings -> conclusions]" -o figures/graphical_abstract.png

Along with this it produces IMRAD-structured sections, correctly formatted citations, reporting-guideline-compliant methods/results sections, and a minimum-count set of schematic and illustrative figures.

Researchers and academic writers preparing manuscripts for journal submission who need rigorous structure, correct citation formatting, reporting-guideline compliance, and mandatory visual elements rather than a generic writing assistant.