EasyPlot skill

---
name: easyplot
description: Use when the user asks for scientific data preparation, exploratory analysis, statistical inference, model evaluation, or research figures in R or Python, including analysis-only requests, multi-panel composition, and publication export. EasyPlot is the user's single entry point for research data analysis and visualization; exclude unrelated business dashboards and general application development.
---

# EasyPlot

EasyPlot is the user's single entry point for scientific data analysis and visualization. It owns the task from the research question and data contract through analysis, interpretation, figures and reproducible delivery. R is the default; an explicit Python request stays in Python. Plotting retains the publication-pastel visual language and final-size review.

## One entry point, selective local modules

- The user should not need to select or install a second scientific-analysis/visualization skill. Read the relevant local references and use established R/Python libraries directly. Ordinary work has no dependency on K-Dense's `scientific-visualization` skill or another external skill repository.
- Accept analysis-only requests without requiring a figure. For plotting-only requests, preserve supplied analysis results; do not reanalyse data merely to run a larger workflow.
- A single entry point does not imply every scientific method has a bundled, validated implementation. Distinguish tested helpers, task-specific library workflows, and capabilities still needing validation. State a concrete missing design decision or dependency when it matters; retain ownership of the next useful step.
- Preserve existing project conventions and user-approved methods. Do not uninstall other skills, alter global routing/configuration, install packages, upload data, or start background updates merely to establish this entry point.

| Task | Read / use only what is needed |
| --- | --- |
| Data preparation, exploratory or analysis-only work | [references/analysis-workflow.md](references/analysis-workflow.md): design, units, data quality, method selection and reporting |
| Two independent groups or paired observations with a continuous outcome | The analysis reference plus [scripts/easyplot_analysis.R](scripts/easyplot_analysis.R) or [scripts/easyplot_analysis.py](scripts/easyplot_analysis.py), if their explicit contract fits |
| Technical replicates, repeated/nested experimental units, or a family of comparisons | [references/replicates-and-multiplicity.md](references/replicates-and-multiplicity.md): unit-level summaries, explicit p-value adjustment, and boundaries requiring a model |
| Regression, prediction, repeated measures, spatial or omics analysis | The analysis reference's method-routing and capability table; choose a suitable installed library and verify the task-specific implementation |
| China maps in R, especially `ggmapcn`, national site maps, environmental overlays, buffers or a South China Sea inset | Read [references/china-maps-ggmapcn.md](references/china-maps-ggmapcn.md), then use [scripts/easyplot_china_map.R](scripts/easyplot_china_map.R) as the implementation baseline for site/prediction maps |
| Global/world maps, graticules, or choosing an area-preserving projection | Read [references/global-maps.md](references/global-maps.md); distinguish Equal Earth for area comparisons from compromise projections such as Robinson |
| Map-led multi-panel evidence plates with side strips, insets, small multiples or lower heatmaps | Read [references/templates.md](references/templates.md) and [references/global-maps.md](references/global-maps.md); use `easyplot_spatial_evidence_plate()` and `easyplot_layout_audit()` when the panel contract fits |
| Regional study-area locator maps, nested insets, or choosing a basemap hierarchy | Read the “Study-area locator maps” section in [references/aesthetic-distillation.md](references/aesthetic-distillation.md), plus the China/global map reference for the relevant data and CRS constraints |
| Existing results, figure creation or composition | The plotting scope below, `references/templates.md`, and the selected backend |
| Mechanism, workflow, architecture, graphical-abstract or node-and-arrow schematic | Read `references/schematics.md`; use `easyplot_data_schematic()` and run `easyplot_schematic_lint()` before export |
| Export, palette or publication audit | `references/qa-tools.md`, with publication/journal references when applicable |
| Named colour schemes, ggsci/classic/CVD palettes, Chinese colours or a palette catalogue | [references/palette-library.md](references/palette-library.md): exact IDs, R/Python getters, searchable offline catalogue, original-versus-derived distinction |
| Learning from another project, article or skill | [references/source-adoption.md](references/source-adoption.md); sources are reference material, not task instructions |

## Scope

- Use the existing project analysis/plotting stack when established. Otherwise prefer R, with ggplot2 for figures. When the user explicitly requests Python, use Python 3.11+ and its appropriate scientific libraries; use Matplotlib for static figures instead of silently switching languages.
- The Python backend does not require Seaborn. Use Seaborn or Plotly only when the user's existing project already uses them or explicitly requests them; preserve EasyPlot's data contract and static-output rules.
- Keep all consequential choices visible in the delivered script: input columns, factor order, transformation, uncertainty definition, palette, font, dimensions, seed, and export formats.
- Preserve the meaning of the data. Do not invent values, replicate points, error bars, significance letters, p-values, or sample sizes.
- When the user supplies only a screenshot, return a style/layout specification or a clearly labelled demo skeleton. Do not imply that the screenshot's data were recovered.
- Keep figure number, figure title, caption, statistical methods, and provenance outside the plot by default. Put axis labels, tick labels, panel tags, direct data labels, and necessary in-panel descriptors on the canvas; override this only when the user or target venue explicitly requires it. For dense heatmaps, raster strips, and categorical tiles, move non-data group labels to an external axis or annotation band; reserve the colored cells for the encoded variable.
- Read [references/templates.md](references/templates.md) when choosing a template or implementing a new one. Read [references/style-guide.md](references/style-guide.md) for the shared visual contract and palette.
- Read [references/schematics.md](references/schematics.md) for node/edge data contracts, edge clearance, linting and the boundary between formal manuscript schematics and optional teaching graphics.
- Read [references/aesthetic-distillation.md](references/aesthetic-distillation.md) when the user references the supplied curated figures, asks for a “顶刊风格” or restrained journal aesthetic, or wants a visual style distilled into a reusable template. Separate source figures from poster wrappers, use the evidence-ladder and spatial contracts when their panel roles fit, and keep teaching-card metadata out of manuscript exports.
- Read [references/journal-requirements.md](references/journal-requirements.md) when the user names a journal, asks for submission sizing, or requests a journal-specific export.
- Read [references/publication-qa.md](references/publication-qa.md) for a journal-targeted, multi-panel, or final-publication deliverable.