Kami - Open-source AI Document Design System, Delivering Professional Documents Ready for Delivery

What Is Kami

Kami is an open-source AI document design system by tw93, distributed as a Claude Code Skill, specifically solving the problem of AI-generated documents having cookie-cutter layouts and cheap visuals. Through a strict set of print-product constraint rules—warm beige canvas, a single ink-blue accent color, fixed-weight serif fonts, tight line spacing, etc.—the tool enables AI to consistently output professional PDFs ready for delivery. Kami supports 8 document types—one-pager reports, long documents, letters, portfolios, resumes, slides, equity reports, changelogs—and 14 embedded SVG charts, with adaptation for Chinese, English, and Japanese.

Kami's Main Features

8 Document Templates: One-Pager, Long Doc (white papers/research reports), Letter (formal correspondence), Portfolio (project portfolio), Resume, Slides (presentation slides), Equity Report (financial analysis), Changelog.
14 Embedded SVG Charts: architecture diagrams, flowcharts, Gantt charts, etc., injected directly by AI without external images.
Trilingual Typesetting: English and Chinese are first-class; Japanese is supported on a best-effort basis via the CJK path and requires visual QA before delivery.
Brand Configuration: Supports ~/.config/kami/brand.md for persisting personal brand colors, identity, writing habits, etc.

Kami's Technical Principles

Constraint-Driven Design: The tool is a set of hard cross-document rules (8 ironclad rules covering canvas color, accent color, gray tones, fonts, font weight, line spacing, shadows, label backgrounds, etc.); every AI output is constrained by the same rule set, avoiding style drift.
Skill Distribution: It exists in the form of a Claude Code / Codex / Claude Desktop Skill; AI triggers it automatically via natural-language requests, with no slash commands needed.
WeasyPrint Rendering: The backend converts HTML/CSS to PDF; specific workarounds were made for WeasyPrint's rgba() double-layer rectangle bug and more.
Font Strategy: Each language uses only one serif font (Chinese: TsangerJinKai 02, English: Charter, Japanese: YuMincho), loaded locally first, falling back to the jsDelivr CDN when missing.

How to Use Kami

Install the Skill: Open an AI programming tool (such as Claude Code or Codex), search for Kami in the official plugin marketplace and add it; or find the skill management entry in Claude Desktop's settings and directly upload the installation package provided by the author, with no extraction needed.
Describe Requirements in Natural Language: After installation, simply tell the AI what document you want—no need to remember any commands at all.
AI Automatically Applies the Design System: The AI automatically recognizes the requirement, invokes the corresponding Kami template, and generates a PDF according to rules such as warm beige canvas, ink-blue accent color, and professional serif fonts.
Download and Update: The generated document is output directly as a PDF and can be printed or sent immediately. To update the Kami version, re-download the latest installation package and replace the old version on the skill management page.

Kami's Key Information and Usage Requirements

Font Licensing: TsangerJinKai 02 is free for personal use; commercial use requires obtaining a license from tsanger.cn; the remaining fonts are bundled with the system or open-source
GitHub Data：4.4k+ Stars / 221 Forks
Installation Prerequisites: Claude Code, Codex, or Claude Desktop already installed
Chinese Support: Complete, including Cang'er Jinkai 02 font
Offline Capability: Fonts load locally first, enabling offline rendering

Kami's Core Advantages

Say Goodbye to "Generic Gray": A warm beige canvas replaces pure white, making both printing and on-screen reading warmer and more professional.
Restrained Single Accent Color: Only ink blue is used, at no more than 5%, keeping the visual focus clear and unflashy.
No bold Design: Fixed font weights, with hierarchy established through font size and left vertical lines, avoiding the cheap feel of arbitrary bolding.
Print-Grade Line Spacing: Compact headings, dense body text, and moderate reading line spacing reject the loose feel of webpages.
Agent-Friendly: The rules are simple enough for AI to execute reliably, and strict enough that every output can be delivered directly.
Lightweight Updates: The installation package is extremely small; simply upload and replace to update.

Kami's Project Addresses

Official Website：https://kami.tw93.fun/
GitHub Repository：https://github.com/tw93/Kami

Comparison of Kami with Similar Competing Products

Dimension	Kami	Claude Artifacts / Default Output	LaTeX / Overleaf
Ease of Use	Install with one command, driven by natural language	Zero cost, but layout is uncontrollable	Steep learning curve, requires handwritten code
Layout Consistency	Strict constraints across documents, unified style	Style drifts with each session	Highly consistent, but templates require manual maintenance
AI-Native	A constraint language designed specifically for Agents	AI-generated directly, but no design system	Not AI-native, requires manual intervention
Chart Support	14 embedded SVGs, self-contained	Relies on external images or Mermaid	Native support for complex formulas and charts
Delivery Format	PDF (print-grade)	Webpages/interactive components	PDF (academic-grade)
Target Audience	AI users who pursue professional typography and do not want to fuss over design	Quick validation, low requirements for layout	Academic/scientific research users

Kami's Application Scenarios

Job Resumes: Kami can help job seekers generate in one click a professional resume PDF with refined layout and a restrained style that can be printed and submitted directly, completely saying goodbye to the cheap visual feel brought by Word's default templates.
Investment Research Reports: Kami can automatically lay out AI-generated individual stock or industry research reports into print-grade PDFs with embedded charts, giving financial analysis documents a reading experience comparable to professional publications.
Funding BPs: Kami supports quickly producing one-pagers or complete business plan PDFs with unified visuals and clear information hierarchy, helping startup teams establish a professional and trustworthy first impression in front of investors.
Presentation Slides: Kami can automatically convert existing text content into presentation slides PDF with the same design style, eliminating the repetitive work of layout design from scratch and giving technical sharing and keynotes more style.
Business Letters: Kami can lay out formal recommendation letters, partnership invitation letters, and more as PDF documents with a paper-like texture and brand consistency, supporting direct printing or sending as professional email attachments.