Lieflat Charts is a data visualization skill for AI agents that generates polished, interactive HTML charts from structured data.
The tool solves the problem of creating publication-ready data visualizations within AI agent workflows. It operates as a skill compatible with the Agent Skills format, designed for use with moxt, Claude Code, Codex, and other AI agents that support the SKILL.md specification. By default, the tool generates editorial-quality charts; it only produces full-page HTML reports from templates when users explicitly request reports, annual reports, monthly reports, whitepapers, posters, or briefs. The skill establishes a unified visual language through consistent typography, whitespace, line weights, and motion effects.
The tool offers three distinct visual styles suited to different communication needs. Lupi presents an editorial narrative approach using fine lines, dot matrices, granular records, and generous whitespace to emphasize authentic units and details, making it suitable for papers, long-form content, annual reports, and data stories requiring careful reading. Glance uses bold bars, large numbers, color blocks, and clear ordering to aggregate information quickly, allowing readers to grasp highs, lows, changes, and anomalies within seconds, fitting weekly reports, briefings, and dashboards. Basics preserves familiar chart outlines like bar charts, line charts, and ring charts while adding measurable scales, hairline details, and editorial refinement for simpler structures or smaller datasets. The tool also provides standalone interactive visualizations for networks, paths, and multi-segment flows. A monochrome grayscale palette serves as the stable default, with color modes supporting celadon blue, coconut green, and editorial red, with custom color palettes available when users provide specific brand colors. Charts preserve data's authentic units while allowing titles, annotations, sources, and page structure to participate in expression.
The project maintains active development with regular updates to templates and visual styles. The codebase is primarily HTML-based, reflecting its focus on generating web-ready visualizations. Documentation includes bilingual support in Chinese and English, with extensive preview galleries demonstrating the range of available templates and motion effects across different visualization types.
