A Claude Code Plugin for Data Communication and Storytelling
"The greatest value of a picture is when it forces us to notice what we never expected to see."
Edward Tufte
DataInk is a Claude Code plugin containing modular skills for clear, effective data communication.
Each skill captures established practices from data visualization, information design, and narrative communication, packaged into structured SKILL.md workflows that guide Claude through chart design, infographic creation, and narrative construction.
Repository: https://github.com/oh-da/dataink.git
Many data visualizations fail not because the data is incorrect but because the communication is weak.
Common problems include:
- pie charts that distort proportions
- 3D charts that skew values
- dashboards overloaded with visual noise
- presentations that show numbers without explaining why they matter
DataInk addresses these issues by encoding expert workflows into reusable AI skills.
These workflows guide an AI system through the steps required to design charts, construct narratives, and communicate insights clearly.
The skills can be used independently or combined into a full communication workflow.
flowchart LR
A[Raw Data or Analysis] --> B[Data Visualization Expert]
B --> C[Clear Individual Charts]
C --> D[Infographic Creator]
D --> E[Structured Visual Summary]
E --> F[Data Storyteller]
F --> G[Presentation or Report]
style B fill:#d4f4dd
style D fill:#d4e6f4
style F fill:#f4e1d4
Transforms raw data into clear charts.
Examples:
- choose appropriate chart type
- remove clutter
- apply visual design principles
Output: clear charts.
Combines insights into a visual summary.
Examples:
- infographic layout
- highlighting key data
- visual storytelling
Output: structured infographic.
Builds a narrative around the visuals.
Examples:
- presentation structure
- narrative arc
- decision framing
Output: presentation or report.
DataInk is useful whenever data must be communicated clearly.
Examples:
- quarterly performance presentations
- strategy updates
- board reports
- KPI dashboards
Recommended workflow:
data-visualization → data-storyteller
Examples:
- redesign cluttered dashboards
- choose the correct chart type
- remove visual noise
Recommended workflow:
data-visualization
Examples:
- product analytics presentations
- marketing performance reviews
- operations reports
- research presentations
Recommended workflow:
data-storyteller
Examples:
- annual reports
- research summaries
- educational visualizations
- policy reports
Recommended workflow:
infographic-creator
DataInk is built on the idea that data visualization is a communication discipline.
Charts often fail because they prioritize visual decoration instead of clarity and narrative structure. Effective data communication requires understanding the audience, identifying the key insight, and presenting information in a way that guides attention.
The workflows in this repository translate these principles into repeatable steps that an AI system can follow.
Before selecting a chart type, it is important to answer:
- Who is the audience
- What decision must be made
- What insight matters most
For this reason the workflows begin with narrative framing.
Exploratory analysis helps analysts discover insights.
Explanatory analysis communicates those insights to others.
The workflows in this repository focus on explanatory communication by removing unnecessary analysis and highlighting the insights that matter.
Effective charts:
- maximize the data ink ratio
- remove unnecessary visual elements
- highlight only a small portion of the visual
- guide the viewer's attention
Data becomes persuasive when presented as part of a narrative.
The storytelling workflows follow a simple structure:
- Setup
- Conflict
- Resolution
Each skill focuses on a specific communication task.
- data-visualization helps design charts
- infographic-creator helps build visual summaries
- data-storyteller helps structure narratives
These skills can be used independently or combined.
DataInk is structured as a Claude Code plugin. Each skill lives under the skills/ directory with optional assets and references folders.
dataink/
├── .claude-plugin/
│ └── plugin.json
└── skills/
├── data-visualization/
│ ├── SKILL.md
│ ├── assets/
│ │ ├── brand-colors.md
│ │ └── brand-fonts.md
│ └── references/
│ ├── medium-selection.md
│ ├── chart-types.md
│ └── design-principles.md
├── data-storyteller/
│ ├── SKILL.md
│ └── references/
│ ├── narrative-arc.md
│ ├── flow-and-repetition.md
│ └── logic-validation.md
└── infographic-creator/
├── SKILL.md
└── references/
├── audience-context.md
└── visual-hierarchy.md
flowchart TD
A[AI Agent] --> B[SKILL.md Instructions]
B --> C[Workflow Steps]
B --> D[Assets]
B --> E[References]
D --> F[Brand Colors]
D --> G[Brand Fonts]
E --> H[Visualization Principles]
E --> I[Storytelling Methods]
The AI loads instructions first and consults assets or references only when necessary.
In Claude Code, run:
/plugin marketplace add oh-da/dataink
/plugin install dataink@dataink
git clone https://github.com/oh-da/dataink.git
claude --plugin-dir ./datainkOnce installed, the following skills are available:
/dataink:visualize— design effective charts/dataink:story— structure data narratives/dataink:infographic— create visual summaries
Claude will also invoke these skills automatically based on task context.
Edit the brand asset files to match your organization:
skills/data-visualization/assets/brand-colors.mdskills/data-visualization/assets/brand-fonts.md
Future additions may include:
- dashboard layout guidelines
- KPI prioritization
- interaction design
- annotation strategies
- insight callouts
- explanation patterns
- color blind safe palettes
- contrast validation
- font readability
A skill that critiques charts and suggests improvements.
Contributions are welcome.
To add a new skill:
- Create a directory under
skills/ - Add a
SKILL.mdwith frontmatter (name,description,metadata) - Include references if needed
- Add assets for brand customization
See the LICENSE file for details.
This repository adapts concepts, frameworks, and principles from the following authors and works. It is intended as a practical workflow tool, not a substitute for reading the original books — which are highly recommended.
The data-storyteller and infographic-creator skills are substantially adapted from techniques described in this book. Specific frameworks and concepts used include:
- The "3-Minute Story" — a planning exercise for distilling the core narrative (Chapter 1)
- The "Big Idea" — a single-sentence formulation of the key message, originally from Nancy Duarte's Resonate (Wiley, 2010) and presented by Knaflic with three structural criteria (Chapter 1)
- The Three-Act Narrative Arc applied to data presentations, with the "setting / main character / imbalance" framing credited by Knaflic to Cliff Atkinson's Beyond Bullet Points (Microsoft Press, 2005) (Chapter 7)
- The tension between "what is" and "what could be" — a framing attributed by Knaflic to Nancy Duarte's Resonate (Chapter 7)
- "Lead with the Ending" — a narrative flow strategy for time-constrained audiences (Chapter 8)
- "Bing, Bang, Bongo" — a three-pass repetition mnemonic (Chapter 8)
- "Repeatable Sound Bites" — attributed by Knaflic to Nancy Duarte (Chapter 8)
- Horizontal Logic, Vertical Logic, and Reverse Storyboarding — presentation validation tactics (Chapter 8)
- The "where are your eyes drawn?" test — a validation exercise for preattentive attribute effectiveness (Chapter 5)
- Action titles — the practice of using insight-driven slide titles rather than topic labels
- Chart selection guidance — mapping data relationships to chart types, including specific recommendations against pie charts, 3D graphics, and secondary y-axes (Chapter 2)
- Application of Gestalt principles (proximity, similarity, closure, enclosure) to chart design (Chapter 3)
- Table vs. graph selection guidance (Chapter 2)
The data-visualization skill also draws on several of these design and chart selection principles.
The seven-category relationship taxonomy used to organize chart selection in the data-visualization skill — time-series, nominal comparison, ranking, part-to-whole, deviation, distribution, and correlation — originates in this work. The color strategy of using muted, natural tones for baseline data and reserving vivid color for emphasis also draws on Few's guidance.
The "data-ink ratio" concept — the principle that the share of a graphic's ink devoted to data should be maximized — originates in this work. The related practices of removing chart borders, background fills, and decorative gridlines are applications of this principle.
The 10% highlighting guideline — the recommendation that at most 10% of a visual's surface should be highlighted — originates in this book. It is cited by Knaflic in Storytelling with Data and is used throughout this repository's skills.
The Gestalt principles of visual perception (proximity, similarity, closure, continuity) originate in early 20th-century perceptual psychology research. Their application to data visualization is widely discussed in the field.
Disclaimer: This repository adapts and reorganizes concepts from the works listed above into structured AI workflows. All borrowed frameworks are attributed to their originators. The text has been written in the author's own words and is not a reproduction of the original works. For the full depth of these ideas, readers are encouraged to consult the source books directly.