Text to Infographic

text-to-infographic v0.1.0

Purpose

Turn user-provided workflows, frameworks, analysis notes, plans, dashboards, or knowledge summaries into a single-page infographic system that is optimized for:

• information hierarchy
• correctness
• scanability
• editability
• cross-tool handoff

This skill is intended for overview visuals that will later be embedded into:

• Lark docs
• Lark sheets with inserted whiteboards
• whiteboard canvases
• SVG-based design workflows
• companion docs that hold the detailed explanation

Best fit

Use this skill when the user wants to:

• explain a process, framework, or decision path in one page
• summarize a complex topic into an overview graphic
• diagnose a problem with cause/effect structure
• compare options or show relationships between entities
• present a roadmap or dashboard that must remain editable later
• prepare a visual artifact first, then let a document carry the details

Common chart families:

• flywheel
• fishbone
• value pyramid
• sankey-like flow
• roadmap
• SaaS dashboard
• process map
• comparison chart

Not a fit

Do not use this skill when the task is mainly:

• multi-panel storytelling
• character-driven continuity
• joke timing or punchline rhythm
• comic payoff
• picture-book page sequencing

Those should remain in text-to-comic.

Core priorities

Always optimize in this order:

1. information hierarchy
2. correctness
3. scanability
4. editability
5. aesthetics

Do not let decorative styling make the diagram harder to understand.

Output contract

Represent the plan with schemas/infographic-plan.schema.json.

The plan should express:

• top-level message
• layout intent
• visual system intent
• content blocks
• relations between blocks
• delivery targets

Keep the plan tool-agnostic:

• do not store absolute x/y coordinates
• do not store whiteboard node IDs
• do not lock the plan to one rendering engine
• let downstream adapters compute actual placement

Default chart mapping

Use these defaults unless the user clearly asks otherwise:

• flywheel -> radial + polar + clockwise
• fishbone -> spine-branch + cartesian + left-to-right
• pyramid -> pyramid + cartesian + top-to-bottom
• roadmap -> timeline + cartesian + left-to-right
• dashboard -> dashboard + cartesian + left-to-right

Default delivery mapping:

• primary_target = whiteboard
• secondary_targets = ["svg", "doc"]
• doc_mode = companion-detail

Workflow

1. Analyze the request

Identify:

• the user’s main message
• the audience
• the chart family that best fits the structure
• whether the output is for explanation, summary, comparison, diagnosis, planning, persuasion, or dashboarding

If multiple chart families are plausible, present the best default and one backup option.

2. Build the infographic plan

Create:

• message
• layout
• visual_system
• blocks
• relations
• delivery

Rules:

• every block must have a stable block_id
• keep text short enough to scan quickly
• use payload for chart-family-specific details when needed
• express relationships explicitly with relations

3. Validate the plan

Use scripts/validate_infographic_plan.py.

Minimum checks:

• schema subset compliance
• unique block_id
• relation endpoints resolve to existing blocks
• sankey relations include weight
• text stays within reasonable density
• default layout and delivery mapping can be applied cleanly

4. Produce adapter drafts

Use scripts/build_infographic_adapters.py.

The adapter layer should emit at least:

• SVG draft structure
• whiteboard draft structure
• doc summary / outline

When the environment exposes tool-specific capabilities such as lark-cli, @larksuite/whiteboard-cli, SVG workflows, or Feishu doc APIs, use the validated plan as the source of truth and map it into those tools.

5. Deliver the artifact

Prefer a compact deliverable bundle:

• validated infographic plan JSON
• adapter drafts
• a short doc summary
• any user-facing notes about assumptions or omitted detail

Hard constraints

Readability over decoration

Prefer fewer blocks with stronger grouping over dense pages full of tiny nodes.

Companion document strategy

The infographic should explain the whole picture quickly. Detailed rationale, definitions, and edge cases should go into the companion doc.

No fake precision

If exact numbers or structure are missing, keep the visual honest:

• label assumptions clearly
• avoid invented metrics
• avoid implying precise proportions without data

Sankey rule

When chart_family = sankey, every relation must include a numeric weight.

Files in this package

• schemas/infographic-plan.schema.json
• examples/*.json
• scripts/validate_infographic_plan.py
• scripts/build_infographic_adapters.py

Quick commands

Validate all examples:

python3 scripts/validate_infographic_plan.py examples/*.json --pretty

Build adapter drafts for one example:

python3 scripts/build_infographic_adapters.py \
  examples/infographic-flywheel-demo.json \
  --pretty