{"owner":"tw93","repo":"Kami","hasSkills":true,"totalSkillsCount":111,"totalTokensCount":59903,"categories":["marketplace","root-instruction","anthropic-skill","claude-rule","plugin-manifest"],"hasMcp":false,"mcpConfig":null,"found":[".claude-plugin/marketplace.json","llms.txt","skills/kami/SKILL.md","plugins/kami/skills/kami/SKILL.md",".agents/plugins/marketplace.json","evals/equity-report-charts-en/prompt.md","site/llms.txt","AGENTS.md","CLAUDE.md","evals/landing-page-en/prompt.md","evals/letter-ko/prompt.md","evals/long-doc-cn/prompt.md","evals/one-pager-en/prompt.md","evals/portfolio-cn/prompt.md","evals/resume-cn/prompt.md","evals/slides-en/prompt.md","skills/kami/CHEATSHEET.md","evals/equity-report-charts-en/graders/content-fidelity.md","evals/equity-report-charts-en/graders/no-placeholders.md","evals/equity-report-charts-en/graders/pdf-created.md","evals/equity-report-charts-en/graders/skill-fired.md","evals/equity-report-cn/graders/content-fidelity.md","evals/equity-report-cn/graders/no-placeholders.md","evals/equity-report-cn/graders/pdf-created.md","evals/equity-report-cn/graders/skill-fired.md","evals/landing-page-en/graders/content-fidelity.md","evals/landing-page-en/graders/html-created.md","evals/landing-page-en/graders/no-placeholders.md","evals/landing-page-en/graders/skill-fired.md","evals/letter-ko/graders/content-fidelity.md","evals/letter-ko/graders/no-placeholders.md","evals/letter-ko/graders/pdf-created.md","evals/letter-ko/graders/skill-fired.md","evals/long-doc-cn/graders/content-fidelity.md","evals/long-doc-cn/graders/no-placeholders.md","evals/long-doc-cn/graders/pdf-created.md","evals/long-doc-cn/graders/skill-fired.md","evals/one-pager-en/graders/content-fidelity.md","evals/one-pager-en/graders/no-invented-metrics.md","evals/one-pager-en/graders/no-placeholders.md","evals/one-pager-en/graders/pdf-created.md","evals/one-pager-en/graders/skill-fired.md","evals/portfolio-cn/graders/content-fidelity.md","evals/portfolio-cn/graders/no-placeholders.md","evals/portfolio-cn/graders/pdf-created.md","evals/portfolio-cn/graders/skill-fired.md","evals/resume-cn/graders/content-fidelity.md","evals/resume-cn/graders/no-placeholders.md","evals/resume-cn/graders/pdf-created.md","evals/resume-cn/graders/skill-fired.md","evals/slides-en/graders/content-fidelity.md","evals/slides-en/graders/no-placeholders.md","evals/slides-en/graders/pdf-created.md","evals/slides-en/graders/skill-fired.md","plugins/kami/.claude-plugin/plugin.json","plugins/kami/.codex-plugin/plugin.json","skills/kami/references/anti-patterns.md","skills/kami/references/brand-profile.md","skills/kami/references/brand.example.md","skills/kami/references/checks_thresholds.json","skills/kami/references/deck-preflight.md","skills/kami/references/design.md","skills/kami/references/diagrams.md","skills/kami/references/mermaid-theme.json","skills/kami/references/mermaid.md","skills/kami/references/production.md","skills/kami/references/resume-writing.md","skills/kami/references/tokens.json","skills/kami/references/writing.md","skills/kami/scripts/package.json","plugins/kami/skills/kami/CHEATSHEET.md","skills/kami/references/schemas/changelog.json","skills/kami/references/schemas/equity-report.json","skills/kami/references/schemas/landing-page.json","skills/kami/references/schemas/letter.json","skills/kami/references/schemas/long-doc.json","skills/kami/references/schemas/one-pager.json","skills/kami/references/schemas/portfolio.json","skills/kami/references/schemas/resume.json","skills/kami/references/schemas/slides.json","skills/kami/scripts/mathjax-runtime/package-lock.json","skills/kami/scripts/mathjax-runtime/package.json","plugins/kami/skills/kami/references/anti-patterns.md","plugins/kami/skills/kami/references/brand-profile.md","plugins/kami/skills/kami/references/brand.example.md","plugins/kami/skills/kami/references/checks_thresholds.json","plugins/kami/skills/kami/references/deck-preflight.md","plugins/kami/skills/kami/references/design.md","plugins/kami/skills/kami/references/diagrams.md","plugins/kami/skills/kami/references/mermaid-theme.json","plugins/kami/skills/kami/references/mermaid.md","plugins/kami/skills/kami/references/production.md","plugins/kami/skills/kami/references/resume-writing.md","plugins/kami/skills/kami/references/tokens.json","plugins/kami/skills/kami/references/writing.md","plugins/kami/skills/kami/scripts/package.json","skills/kami/assets/templates/marp/slides-marp-en.md","skills/kami/assets/templates/marp/slides-marp.md","plugins/kami/skills/kami/references/schemas/changelog.json","plugins/kami/skills/kami/references/schemas/equity-report.json","plugins/kami/skills/kami/references/schemas/landing-page.json","plugins/kami/skills/kami/references/schemas/letter.json","plugins/kami/skills/kami/references/schemas/long-doc.json","plugins/kami/skills/kami/references/schemas/one-pager.json","plugins/kami/skills/kami/references/schemas/portfolio.json","plugins/kami/skills/kami/references/schemas/resume.json","plugins/kami/skills/kami/references/schemas/slides.json","plugins/kami/skills/kami/scripts/mathjax-runtime/package-lock.json","plugins/kami/skills/kami/scripts/mathjax-runtime/package.json","plugins/kami/skills/kami/assets/templates/marp/slides-marp-en.md","plugins/kami/skills/kami/assets/templates/marp/slides-marp.md"],"skills":{".claude-plugin/marketplace.json":"{\n \"name\": \"kami\",\n \"description\": \"Document typesetting skill for Claude Code.\",\n \"owner\": {\n \"name\": \"Tw93\",\n \"email\": \"hitw93@gmail.com\"\n },\n \"plugins\": [\n {\n \"name\": \"kami\",\n \"version\": \"1.18.0\",\n \"description\": \"Typeset professional documents and landing pages with the Kami design system.\",\n \"category\": \"documents\",\n \"source\": \"./plugins/kami\",\n \"homepage\": \"https://github.com/tw93/kami\"\n }\n ]\n}\n","llms.txt":"site/llms.txt","skills/kami/SKILL.md":"---\nname: kami\ndescription: 'Typeset professional documents with Kami templates: resumes, one-pagers, white papers, letters, portfolios, and slide decks. Use when asked to 做 PDF / 排版 / 简历 / 一页纸 / PPT / slides, or to create a Kami landing page. Not for auditing or restyling an existing product site.'\n---\n\n# kami · 紙\n\n**紙 · かみ** means paper in Japanese.\n\nGood content deserves good paper. Use Kami’s templates and layout rules to create documents and landing pages with serif fonts, warm backgrounds, and ink-blue accents.\n\nPart of `Kaku · Waza · Kami` - Kaku writes code, Waza drills habits, **Kami delivers documents**.\n\nCommands and paths below are relative to this skill's directory. Keep the document in the user's working directory and call the scripts from there by their full path (`python3 /scripts/build.py ...`); relative file arguments resolve from your working directory first.\n\n**Update check (non-blocking).** At the start of a task, run `bash scripts/check-update.sh`. It keeps a local daily marker and, at most once a day, reads Kami's latest published GitHub Release; it uploads no document or task content and stays silent when offline or sandboxed. When it prints a line about a newer Kami, relay it and continue.\n\n**Brand profile.** If `~/.config/kami/brand.md` (or the legacy `~/.kami/brand.md`) exists, read `references/brand-profile.md` and apply it. Precedence: explicit prompt > editorial judgment > habit notes > frontmatter defaults > built-in defaults. The profile fills gaps silently and never overrides the current conversation. When the user names another project as the visual reference, run the style scan in the same file. When they give only a brand website, `python3 scripts/brand_extract.py ` drafts a profile; show the draft and use it for this document only after they accept it.\n\n## 1 · Lock the contract\n\nBefore creating or changing an output, lock six things: language, template, output formats, page or length target, visual acceptance, and the verification command. Infer them from the request, then state the contract to the user in one short line in their language (template, length, formats, the figures you plan, any missing material) and continue without waiting. If they push back, adjust.\n\n**Question budget.** Infer first. Ask at most once, in one compact question with at most two parts, and only when two or more of purpose, audience, hard constraints, or success are genuinely unresolvable, or two templates genuinely both fit. Otherwise take the smallest reasonable assumption and report it at the end.\n\nRoute by the artifact's current state:\n\n| Current task | Mode | Contract |\n|---|---|---|\n| New document or substantial restructuring | **New document** | Lock the contract, write `content.json` with `brief` + `content`, then fill and deliver |\n| Text replacement, translation, or factual correction in an existing artifact | **Content-only** | Preserve CSS and layout unless the new copy proves a fit defect |\n| User supplies a render or screenshot and rejects how it looks | **Visual repair** | Treat the render as the brief; follow «Feedback protocol» |\n| Standalone generated illustration, cover, social card, or redraw | **Generated asset** | Lock the semantic brief before pixels (`references/diagrams.md` section 11); preserve accepted parts across iterations |\n\nUse the nearest existing template and check. Do not add a template, shared CSS layer, dependency, script flag, or optional mode unless the request cannot be met without it.\n\n### Language\n\n| User language | HTML templates | Slides (PDF default) | Slides (PPTX, only when an editable deck is requested) | Slides (Marp) |\n|---|---|---|---|---|\n| Chinese (primary) | `*.html` | `slides-weasy.html` | `slides.py` | `slides-marp.md` |\n| English | `*-en.html` | `slides-weasy-en.html` | `slides-en.py` | `slides-marp-en.md` |\n| Japanese (best-effort) | `*.html`, JP Mincho first | `slides-weasy.html` | `slides.py` | `slides-marp.md` |\n| Korean | `*-ko.html` | `slides-weasy-ko.html` | `slides-en.py` only if PPTX is required | `slides-marp.md` |\n| Other (best-effort) | CJK or EN path by script coverage | `slides-weasy.html` or `-en` | `slides.py` / `slides-en.py` | `slides-marp.md` or `-en` |\n\nJapanese, Korean, and other languages need a visual pass on the page images before shipping. Reference docs are shared English specs.\n\n### Document type\n\n| User says | Document | CN template | EN template | KO template |\n|---|---|---|---|---|\n| \"one-pager / 方案 / 执行摘要 / exec summary\" | One-Pager | `one-pager.html` | `one-pager-en.html` | `one-pager-ko.html` |\n| \"white paper / 白皮书 / 长文 / 年度总结 / technical report\" | Long Doc | `long-doc.html` | `long-doc-en.html` | `long-doc-ko.html` |\n| \"formal letter / 信件 / 辞职信 / 推荐信 / memo\" | Letter | `letter.html` | `letter-en.html` | `letter-ko.html` |\n| \"portfolio / 作品集 / case studies\" | Portfolio | `portfolio.html` | `portfolio-en.html` | `portfolio-ko.html` |\n| \"resume / CV / 简历 / 履歴書\" | Resume | `resume.html` | `resume-en.html` | `resume-ko.html` |\n| \"slides / PPT / deck / 演示\" | Slides | `slides-weasy.html` | `slides-weasy-en.html` | `slides-weasy-ko.html` |\n| \"个股研报 / equity report / 估值分析 / investment memo / 股票分析\" | Equity Report | `equity-report.html` | `equity-report-en.html` | `equity-report-ko.html` |\n| \"更新日志 / changelog / release notes / 版本记录\" | Changelog | `changelog.html` | `changelog-en.html` | `changelog-ko.html` |\n| \"landing page / 落地页 / 官网 / product page / 产品页\" | Landing Page | `landing-page.html` | `landing-page-en.html` | `landing-page-ko.html` |\n\nWhen two cells both fit, decide from the signals before asking:\n\n| Signal | Document |\n|---|---|\n| 1 page for investors, recruiters, or an executive summary | one-pager |\n| 1 page of formal correspondence (sales, hiring, resignation, memo) | letter |\n| Exactly 2 pages of career narrative with project bullets | resume |\n| 3-6 visual-heavy pages showcasing projects | portfolio |\n| 6-15 pages of sustained argument, low visual density | long-doc |\n| Presentation flow, one assertion per slide | slides |\n| Metrics dashboard plus thesis plus price or risk view | equity-report |\n| Version-by-version release facts | changelog |\n| Product showcase, pricing, screenshots, FAQ for a browser | landing-page |\n\nAsk only in cases like \"1.5 pages of career story, heavy visuals\" (resume or portfolio?) or \"2 pages of exec summary with metric tiles\" (one-pager or equity-report?).\n\n- **Changelog vs. release notes**: the changelog template is a styled document. GitHub release notes are a separate deliverable; use `/write`.\n- **Landing page**: screen-first HTML, no PDF. Fill the `{{PLACEHOLDER}}` values and comment blocks and save a ready-to-serve file. The accent follows the product: use its own main color, ask the user to pick when it has none, and keep ink-blue only as the fallback (`references/design.md` Section 11 «Accent on a product site»). Companion files (`landing-page-*.example`) and the production site mode (docs, help, releases, legal pages, or more than two locales) live in `references/design.md` Section 11. Keep project-specific release artifacts, payment providers, appcast rules, and private local paths out of Kami.\n- **Diagrams inside a document** come from `assets/diagrams/` (eighteen types), not a template. `references/diagrams.md` section 1 picks the type; extract the `` block into a `
`. The same file owns architecture boards, diagrams maintained in the user's repo, and generated raster illustrations.\n- **Charts from data** (pick and embed them without asking) are generated, never drawn by hand: write the numbers as a spec and run `python3 scripts/build.py --chart spec.json -o chart.svg` (`references/diagrams.md` section 9).\n- **Slides**: read `references/deck-preflight.md` before drafting.\n\n## 2 · Sources, materials, and content\n\n**Source check.** When the document names a company, product, person, release date, version, funding round, metric, market fact, or technical spec, work from primary sources, note source and date for the facts that drive it, and ask when sources conflict (`references/writing.md` «5. Sources before phrasing»). Skip it for personal drafts where the user supplied everything.\n\n**Material check.** A branded subject needs what makes it recognizable: logo, product image or current UI screenshot, and brand colors when the document is branded (`references/writing.md` «6. Materials serve recognition»). When the request names no logo but the brand profile has one, fill the commented `.brand-logo` slot in `one-pager` / `portfolio` / `slides-weasy` (expand `~` to an absolute path; leave the slot commented if the file is missing). An explicit logo in the current request always wins.\n\n**Gap report.** Everything the template needs that the input lacks, missing facts and missing materials alike, goes into one compact table shown once, with one question only when a missing item would change the deliverable. Inside the document, mark a gap as `[DATA NEEDED: what]`. Never fill a gap with a guess, a stock image, an approximated logo, or an invented value.\n\n**Distill.** Unless the content already maps one-to-one onto the template sections or the user said to use it as-is, distill it. Every factual claim, number, date, name, source, and material reference from the input must land in a template section or in the gap report.\n\n**Content IR (new documents).** Write the distilled content and the locked contract to `content.json` next to the working HTML before filling. Read `references/schemas/.json` first: its `$comment` notes carry the per-field quality bar. The envelope is strict: only `type`, `lang` (`cn`, `en`, `ko`, `zh-TW`, ...), `brief`, and `content`.\n\n```json\n{\n \"type\": \"resume\",\n \"lang\": \"cn\",\n \"brief\": {\n \"audience\": \"Hiring manager for a senior product role\",\n \"job\": \"Earn an interview by proving scope and outcomes\",\n \"template\": \"resume\",\n \"formats\": [\"html\", \"pdf\"],\n \"page_target\": 2,\n \"narrative\": \"Scope first, then evidence, then fit\",\n \"required_facts\": [\"team size\", \"measured outcomes\"],\n \"required_assets\": [],\n \"acceptance_checks\": [\"two pages\", \"all atomic facts survive\"],\n \"preserve\": [],\n \"explicit_deviations\": []\n },\n \"content\": { ... }\n}\n```\n\n`type` is a schema name in `references/schemas/` (one-pager, letter, resume, long-doc, portfolio, slides, equity-report, changelog, landing-page). For a visual repair, add `target`, `evidence`, and `preserve` to `brief`. Validate before layout with `python3 scripts/build.py --check-content content.json` and fix errors by fixing the content (or asking for the fact), never by loosening structure. Short atomic values (names, metrics, dates, bullets) must survive into the document verbatim; values over 80 characters may be rephrased.\n\n## 3 · Read the right amount\n\nRead the lowest tier that covers the task. The templates already carry every token and component, so a new document does not need the full design spec.\n\n| Task | Read |\n|---|---|\n| Content-only change | `CHEATSHEET.md` |\n| Layout tweak within spec | `CHEATSHEET.md` + the template |\n| New document | the template + `CHEATSHEET.md` + `references/schemas/.json` + the `references/writing.md` sections for that type («Per-document strategies», «Quality bars by document type»); open `references/design.md` only for the component or rule you are about to change |\n| Resume | + `references/resume-writing.md` |\n| Slides, long-doc, portfolio, equity-report, changelog | + `references/writing.md` «Page density» |\n| Slides | + `references/deck-preflight.md` and the `references/design.md` section 8 «Deck Recipe» it points to |\n| Diagram | `references/diagrams.md`; `references/mermaid.md` when the source is Mermaid text |\n| Company, product, market, or branded subject | `references/writing.md` sections 5-6 |\n| Rendering bug, font issue, overflow | `references/production.md` Part 4 (and the design spec if CSS is the cause) |\n| Reviewing an AI-generated draft | `references/anti-patterns.md` |\n\n## 4 · Fill the template\n\n- Copy the template into the working directory; never write HTML from scratch. CSS stays untouched during a content fill. A layout change stays within spec, and a real style change moves `references/design.md` and the sibling templates together, never one file.\n- Meet the quality bar for the type in `references/writing.md` «Quality bars by document type»: a resume bullet needs action, honest scope, and an observable result; an equity report needs variant perception and quantified catalysts; slides need assertion titles. Structure is necessary, not sufficient.\n- Do not invent metrics, financial data, or statistics, and keep any figure you compute from the supplied ones next to its inputs or marked as derived; do not describe stock images as placeholders; do not pad slots (three real projects do not become five). Avoid the patterns in `references/anti-patterns.md`.\n- Tables separate rows with whitespace first and neutral hairlines second; decoration earns its place only by encoding data, state, grouping, or a relationship (`references/design.md` «Table (kami-table)» and «Subtractive rule»).\n- Resume: run the recruiter pass in `references/resume-writing.md` before building.\n- Mathematics is strict LaTeX: author formulas only as inline `\\( ... \\)` or display `\\[ ... \\]` in rendered body text, never as Unicode approximations, ASCII fractions, screenshots, or raw TeX on the page. `--deliver` renders them to MathJax SVG in place; it needs Node.js 20 or 22+ and `bash scripts/ensure_mathjax.sh` once.\n- Code blocks with `class=\"language-*\"` are highlighted when optional `Pygments` is installed; without it they render monochrome.\n\nFill the four metadata placeholders in `
`; WeasyPrint writes them into the PDF:\n\n| CN | EN | Rule |\n|---|---|---|\n| `{{作者}}` | `{{AUTHOR}}` | Resume, letter, portfolio: the person's name from the document. Others: leave it, the build infers it from `git config user.name`, then `KAMI_AUTHOR` |\n| `{{摘要}}` | `{{DESCRIPTION}}` | One sentence (≤150 chars) from the first two paragraphs |\n| `{{关键词}}` | `{{KEYWORDS}}` | 3-5 keywords from the title and section headings |\n| `{{文档标题}}` etc. | `{{DOC_TITLE}}` etc. | From the H1 or `.header .title` |\n\nKeep `` as is.\n\n**Output formats.** Do not ask; decide from context. Any document: HTML + PDF (PDF is the deliverable, HTML the source). Slides: HTML + PDF, plus PPTX only when an editable deck is requested. \"分享 / 发朋友圈 / share / post / preview\": add PNG. \"嵌入 / 插图 / embed in another doc\": PNG only. Landing page: the static HTML file. Word or `.docx` for long-doc, proposal, or equity-report: the PDF plus a DOCX through `references/production.md` Part 5. An explicit format request overrides all of this.\n\n## 5 · Deliver\n\nOne command finishes a document. Run it from the document's directory, fix every `ERROR`, and re-run until it reports `READY`:\n\n```bash\npython3 scripts/build.py --deliver filled.html content.json\n```\n\nIt checks the HTML (placeholders, strict math rendered in place, Markdown residue, generated charts against their data, template style, content coverage), renders the PDF next to it, then checks the PDF (page contract, resume balance, the font that actually drew CJK text, residue, density, orphans) and exports page images. `WARN` lines point at where to look, not at failures. `GAP` lines are the `[DATA NEEDED]` markers you must report.\n\nThen do the part no script can: open every page image and walk the printed checklist. One hit means a whole-document sweep for that class of issue. Fix the three most visible defects per round, re-run `--deliver`, and stop after three rounds; list what is still wrong rather than looping. If the host cannot view images, send the image paths and the checklist to the user instead of skipping the pass. A missing CJK serif shows no fallback boxes, only a heavier, flatter page, which is why the font gate exists; never call a CJK document visually verified without it.\n\nOther surfaces:\n\n- **Landing pages and any browser surface**: `--deliver` checks the HTML only. Screenshot the responsive matrix in every shipped locale before calling it done (`references/design.md` «Responsive screenshot verification»); do not wait for the user to ask about mobile.\n- **Editable PPTX**: `python3 scripts/build.py --verify slides` (or `slides-en`) and `--check-rhythm`.\n- **Missing dependencies or fonts**: `python3 scripts/build.py --doctor` reports what is installed; `bash scripts/ensure-fonts.sh` recovers CJK fonts (the commercial TsangerJinKai02 never ships inside the skill package). Render failures and their fixes: `references/production.md` Part 4.\n\n**Fresh review.** After `READY` and the image pass, review once from the contract rather than from your own rationale: reread `brief`, the coverage result, and the page images; check every acceptance item and `preserve` boundary; fix P0/P1 findings, each tied to the page or element that proves it. Use an isolated reviewer when the host has one.\n\n**Definition of done.** The closing message gives:\n\n1. The path of every deliverable, in every promised format.\n2. The `--deliver` verdict and page count against the contract.\n3. Every `[DATA NEEDED]` gap and every assumption you made, listed.\n4. The visual verdict stated honestly: when page images could not be inspected, say \"build verified, visuals unconfirmed\", not \"done\".\n\n## Feedback protocol\n\nWhen the user gives visual feedback (\"looks off\", \"太挤了\", \"not elegant\"), inspect the current render before asking them to choose a value. The render is the evidence; the user's negative label is the acceptance signal.\n\n1. Name the concrete defect in one sentence: page or surface, viewport or state, and whether it is density, hierarchy, alignment, type, color, cropping, or text fit.\n2. Lock the repair boundary: `target` is allowed to change; `preserve` names the adjacent pages, sections, content, and shared tokens that must stay stable. Ask only when two plausible targets would produce materially different artifacts.\n3. Make the smallest content, geometry, spacing, typography, crop, or token change that fixes the defect. Never hide a content problem by shrinking type first.\n4. Verify the affected matrix rather than one screenshot:\n - PDF: target page, neighboring pages, total page count, font result, and every locale or template variant reached by a shared token.\n - Screen: the breakpoint, tablet, and baseline matrix in `references/design.md` «Responsive screenshot verification»; every shipped locale; affected default, focus/selected, loading, empty/error, and transition states only when the surface has them.\n - PPTX: editable source plus a rendered PDF or opened-deck inspection.\n - Generated asset: target slot at its smallest display size plus sibling assets in the same deliverable.\n\nNever say \"I'll adjust the spacing\" without naming the exact property and its new value.\n\n**Escalate after two rounds.** If the same element is still not approved after two adjustment rounds, stop nudging values and produce one comparison artifact: the current state plus 2-3 labeled variants (A/B/C) of the same content in the product's actual frame and background, changing only the compared property. For choices with no objective criterion (typeface, accent color, logo), start there: a specimen sheet of up to 5 labeled half-page blocks with identical title-plus-paragraph content. After the pick, apply it everywhere and rebuild what it touches in the same round.\n\n## When not to use this skill\n\n- The user explicitly wants Material, Fluent, or a Tailwind default: a different design language.\n- Dark, cyberpunk, or futurist aesthetics (this is deliberately anti-future).\n- Saturated multi-color palettes (this has one accent).\n- Cartoon, animation, or illustration style (this is editorial).\n- Dynamic web app UI (this is for print and static documents).\n","plugins/kami/skills/kami/SKILL.md":"---\nname: kami\ndescription: 'Typeset professional documents with Kami templates: resumes, one-pagers, white papers, letters, portfolios, and slide decks. Use when asked to 做 PDF / 排版 / 简历 / 一页纸 / PPT / slides, or to create a Kami landing page. Not for auditing or restyling an existing product site.'\n---\n\n# kami · 紙\n\n**紙 · かみ** means paper in Japanese.\n\nGood content deserves good paper. Use Kami’s templates and layout rules to create documents and landing pages with serif fonts, warm backgrounds, and ink-blue accents.\n\nPart of `Kaku · Waza · Kami` - Kaku writes code, Waza drills habits, **Kami delivers documents**.\n\nCommands and paths below are relative to this skill's directory. Keep the document in the user's working directory and call the scripts from there by their full path (`python3 /scripts/build.py ...`); relative file arguments resolve from your working directory first.\n\n**Update check (non-blocking).** At the start of a task, run `bash scripts/check-update.sh`. It keeps a local daily marker and, at most once a day, reads Kami's latest published GitHub Release; it uploads no document or task content and stays silent when offline or sandboxed. When it prints a line about a newer Kami, relay it and continue.\n\n**Brand profile.** If `~/.config/kami/brand.md` (or the legacy `~/.kami/brand.md`) exists, read `references/brand-profile.md` and apply it. Precedence: explicit prompt > editorial judgment > habit notes > frontmatter defaults > built-in defaults. The profile fills gaps silently and never overrides the current conversation. When the user names another project as the visual reference, run the style scan in the same file. When they give only a brand website, `python3 scripts/brand_extract.py ` drafts a profile; show the draft and use it for this document only after they accept it.\n\n## 1 · Lock the contract\n\nBefore creating or changing an output, lock six things: language, template, output formats, page or length target, visual acceptance, and the verification command. Infer them from the request, then state the contract to the user in one short line in their language (template, length, formats, the figures you plan, any missing material) and continue without waiting. If they push back, adjust.\n\n**Question budget.** Infer first. Ask at most once, in one compact question with at most two parts, and only when two or more of purpose, audience, hard constraints, or success are genuinely unresolvable, or two templates genuinely both fit. Otherwise take the smallest reasonable assumption and report it at the end.\n\nRoute by the artifact's current state:\n\n| Current task | Mode | Contract |\n|---|---|---|\n| New document or substantial restructuring | **New document** | Lock the contract, write `content.json` with `brief` + `content`, then fill and deliver |\n| Text replacement, translation, or factual correction in an existing artifact | **Content-only** | Preserve CSS and layout unless the new copy proves a fit defect |\n| User supplies a render or screenshot and rejects how it looks | **Visual repair** | Treat the render as the brief; follow «Feedback protocol» |\n| Standalone generated illustration, cover, social card, or redraw | **Generated asset** | Lock the semantic brief before pixels (`references/diagrams.md` section 11); preserve accepted parts across iterations |\n\nUse the nearest existing template and check. Do not add a template, shared CSS layer, dependency, script flag, or optional mode unless the request cannot be met without it.\n\n### Language\n\n| User language | HTML templates | Slides (PDF default) | Slides (PPTX, only when an editable deck is requested) | Slides (Marp) |\n|---|---|---|---|---|\n| Chinese (primary) | `*.html` | `slides-weasy.html` | `slides.py` | `slides-marp.md` |\n| English | `*-en.html` | `slides-weasy-en.html` | `slides-en.py` | `slides-marp-en.md` |\n| Japanese (best-effort) | `*.html`, JP Mincho first | `slides-weasy.html` | `slides.py` | `slides-marp.md` |\n| Korean | `*-ko.html` | `slides-weasy-ko.html` | `slides-en.py` only if PPTX is required | `slides-marp.md` |\n| Other (best-effort) | CJK or EN path by script coverage | `slides-weasy.html` or `-en` | `slides.py` / `slides-en.py` | `slides-marp.md` or `-en` |\n\nJapanese, Korean, and other languages need a visual pass on the page images before shipping. Reference docs are shared English specs.\n\n### Document type\n\n| User says | Document | CN template | EN template | KO template |\n|---|---|---|---|---|\n| \"one-pager / 方案 / 执行摘要 / exec summary\" | One-Pager | `one-pager.html` | `one-pager-en.html` | `one-pager-ko.html` |\n| \"white paper / 白皮书 / 长文 / 年度总结 / technical report\" | Long Doc | `long-doc.html` | `long-doc-en.html` | `long-doc-ko.html` |\n| \"formal letter / 信件 / 辞职信 / 推荐信 / memo\" | Letter | `letter.html` | `letter-en.html` | `letter-ko.html` |\n| \"portfolio / 作品集 / case studies\" | Portfolio | `portfolio.html` | `portfolio-en.html` | `portfolio-ko.html` |\n| \"resume / CV / 简历 / 履歴書\" | Resume | `resume.html` | `resume-en.html` | `resume-ko.html` |\n| \"slides / PPT / deck / 演示\" | Slides | `slides-weasy.html` | `slides-weasy-en.html` | `slides-weasy-ko.html` |\n| \"个股研报 / equity report / 估值分析 / investment memo / 股票分析\" | Equity Report | `equity-report.html` | `equity-report-en.html` | `equity-report-ko.html` |\n| \"更新日志 / changelog / release notes / 版本记录\" | Changelog | `changelog.html` | `changelog-en.html` | `changelog-ko.html` |\n| \"landing page / 落地页 / 官网 / product page / 产品页\" | Landing Page | `landing-page.html` | `landing-page-en.html` | `landing-page-ko.html` |\n\nWhen two cells both fit, decide from the signals before asking:\n\n| Signal | Document |\n|---|---|\n| 1 page for investors, recruiters, or an executive summary | one-pager |\n| 1 page of formal correspondence (sales, hiring, resignation, memo) | letter |\n| Exactly 2 pages of career narrative with project bullets | resume |\n| 3-6 visual-heavy pages showcasing projects | portfolio |\n| 6-15 pages of sustained argument, low visual density | long-doc |\n| Presentation flow, one assertion per slide | slides |\n| Metrics dashboard plus thesis plus price or risk view | equity-report |\n| Version-by-version release facts | changelog |\n| Product showcase, pricing, screenshots, FAQ for a browser | landing-page |\n\nAsk only in cases like \"1.5 pages of career story, heavy visuals\" (resume or portfolio?) or \"2 pages of exec summary with metric tiles\" (one-pager or equity-report?).\n\n- **Changelog vs. release notes**: the changelog template is a styled document. GitHub release notes are a separate deliverable; use `/write`.\n- **Landing page**: screen-first HTML, no PDF. Fill the `{{PLACEHOLDER}}` values and comment blocks and save a ready-to-serve file. The accent follows the product: use its own main color, ask the user to pick when it has none, and keep ink-blue only as the fallback (`references/design.md` Section 11 «Accent on a product site»). Companion files (`landing-page-*.example`) and the production site mode (docs, help, releases, legal pages, or more than two locales) live in `references/design.md` Section 11. Keep project-specific release artifacts, payment providers, appcast rules, and private local paths out of Kami.\n- **Diagrams inside a document** come from `assets/diagrams/` (eighteen types), not a template. `references/diagrams.md` section 1 picks the type; extract the `` block into a `
`. The same file owns architecture boards, diagrams maintained in the user's repo, and generated raster illustrations.\n- **Charts from data** (pick and embed them without asking) are generated, never drawn by hand: write the numbers as a spec and run `python3 scripts/build.py --chart spec.json -o chart.svg` (`references/diagrams.md` section 9).\n- **Slides**: read `references/deck-preflight.md` before drafting.\n\n## 2 · Sources, materials, and content\n\n**Source check.** When the document names a company, product, person, release date, version, funding round, metric, market fact, or technical spec, work from primary sources, note source and date for the facts that drive it, and ask when sources conflict (`references/writing.md` «5. Sources before phrasing»). Skip it for personal drafts where the user supplied everything.\n\n**Material check.** A branded subject needs what makes it recognizable: logo, product image or current UI screenshot, and brand colors when the document is branded (`references/writing.md` «6. Materials serve recognition»). When the request names no logo but the brand profile has one, fill the commented `.brand-logo` slot in `one-pager` / `portfolio` / `slides-weasy` (expand `~` to an absolute path; leave the slot commented if the file is missing). An explicit logo in the current request always wins.\n\n**Gap report.** Everything the template needs that the input lacks, missing facts and missing materials alike, goes into one compact table shown once, with one question only when a missing item would change the deliverable. Inside the document, mark a gap as `[DATA NEEDED: what]`. Never fill a gap with a guess, a stock image, an approximated logo, or an invented value.\n\n**Distill.** Unless the content already maps one-to-one onto the template sections or the user said to use it as-is, distill it. Every factual claim, number, date, name, source, and material reference from the input must land in a template section or in the gap report.\n\n**Content IR (new documents).** Write the distilled content and the locked contract to `content.json` next to the working HTML before filling. Read `references/schemas/.json` first: its `$comment` notes carry the per-field quality bar. The envelope is strict: only `type`, `lang` (`cn`, `en`, `ko`, `zh-TW`, ...), `brief`, and `content`.\n\n```json\n{\n \"type\": \"resume\",\n \"lang\": \"cn\",\n \"brief\": {\n \"audience\": \"Hiring manager for a senior product role\",\n \"job\": \"Earn an interview by proving scope and outcomes\",\n \"template\": \"resume\",\n \"formats\": [\"html\", \"pdf\"],\n \"page_target\": 2,\n \"narrative\": \"Scope first, then evidence, then fit\",\n \"required_facts\": [\"team size\", \"measured outcomes\"],\n \"required_assets\": [],\n \"acceptance_checks\": [\"two pages\", \"all atomic facts survive\"],\n \"preserve\": [],\n \"explicit_deviations\": []\n },\n \"content\": { ... }\n}\n```\n\n`type` is a schema name in `references/schemas/` (one-pager, letter, resume, long-doc, portfolio, slides, equity-report, changelog, landing-page). For a visual repair, add `target`, `evidence`, and `preserve` to `brief`. Validate before layout with `python3 scripts/build.py --check-content content.json` and fix errors by fixing the content (or asking for the fact), never by loosening structure. Short atomic values (names, metrics, dates, bullets) must survive into the document verbatim; values over 80 characters may be rephrased.\n\n## 3 · Read the right amount\n\nRead the lowest tier that covers the task. The templates already carry every token and component, so a new document does not need the full design spec.\n\n| Task | Read |\n|---|---|\n| Content-only change | `CHEATSHEET.md` |\n| Layout tweak within spec | `CHEATSHEET.md` + the template |\n| New document | the template + `CHEATSHEET.md` + `references/schemas/.json` + the `references/writing.md` sections for that type («Per-document strategies», «Quality bars by document type»); open `references/design.md` only for the component or rule you are about to change |\n| Resume | + `references/resume-writing.md` |\n| Slides, long-doc, portfolio, equity-report, changelog | + `references/writing.md` «Page density» |\n| Slides | + `references/deck-preflight.md` and the `references/design.md` section 8 «Deck Recipe» it points to |\n| Diagram | `references/diagrams.md`; `references/mermaid.md` when the source is Mermaid text |\n| Company, product, market, or branded subject | `references/writing.md` sections 5-6 |\n| Rendering bug, font issue, overflow | `references/production.md` Part 4 (and the design spec if CSS is the cause) |\n| Reviewing an AI-generated draft | `references/anti-patterns.md` |\n\n## 4 · Fill the template\n\n- Copy the template into the working directory; never write HTML from scratch. CSS stays untouched during a content fill. A layout change stays within spec, and a real style change moves `references/design.md` and the sibling templates together, never one file.\n- Meet the quality bar for the type in `references/writing.md` «Quality bars by document type»: a resume bullet needs action, honest scope, and an observable result; an equity report needs variant perception and quantified catalysts; slides need assertion titles. Structure is necessary, not sufficient.\n- Do not invent metrics, financial data, or statistics, and keep any figure you compute from the supplied ones next to its inputs or marked as derived; do not describe stock images as placeholders; do not pad slots (three real projects do not become five). Avoid the patterns in `references/anti-patterns.md`.\n- Tables separate rows with whitespace first and neutral hairlines second; decoration earns its place only by encoding data, state, grouping, or a relationship (`references/design.md` «Table (kami-table)» and «Subtractive rule»).\n- Resume: run the recruiter pass in `references/resume-writing.md` before building.\n- Mathematics is strict LaTeX: author formulas only as inline `\\( ... \\)` or display `\\[ ... \\]` in rendered body text, never as Unicode approximations, ASCII fractions, screenshots, or raw TeX on the page. `--deliver` renders them to MathJax SVG in place; it needs Node.js 20 or 22+ and `bash scripts/ensure_mathjax.sh` once.\n- Code blocks with `class=\"language-*\"` are highlighted when optional `Pygments` is installed; without it they render monochrome.\n\nFill the four metadata placeholders in `
`; WeasyPrint writes them into the PDF:\n\n| CN | EN | Rule |\n|---|---|---|\n| `{{作者}}` | `{{AUTHOR}}` | Resume, letter, portfolio: the person's name from the document. Others: leave it, the build infers it from `git config user.name`, then `KAMI_AUTHOR` |\n| `{{摘要}}` | `{{DESCRIPTION}}` | One sentence (≤150 chars) from the first two paragraphs |\n| `{{关键词}}` | `{{KEYWORDS}}` | 3-5 keywords from the title and section headings |\n| `{{文档标题}}` etc. | `{{DOC_TITLE}}` etc. | From the H1 or `.header .title` |\n\nKeep `` as is.\n\n**Output formats.** Do not ask; decide from context. Any document: HTML + PDF (PDF is the deliverable, HTML the source). Slides: HTML + PDF, plus PPTX only when an editable deck is requested. \"分享 / 发朋友圈 / share / post / preview\": add PNG. \"嵌入 / 插图 / embed in another doc\": PNG only. Landing page: the static HTML file. Word or `.docx` for long-doc, proposal, or equity-report: the PDF plus a DOCX through `references/production.md` Part 5. An explicit format request overrides all of this.\n\n## 5 · Deliver\n\nOne command finishes a document. Run it from the document's directory, fix every `ERROR`, and re-run until it reports `READY`:\n\n```bash\npython3 scripts/build.py --deliver filled.html content.json\n```\n\nIt checks the HTML (placeholders, strict math rendered in place, Markdown residue, generated charts against their data, template style, content coverage), renders the PDF next to it, then checks the PDF (page contract, resume balance, the font that actually drew CJK text, residue, density, orphans) and exports page images. `WARN` lines point at where to look, not at failures. `GAP` lines are the `[DATA NEEDED]` markers you must report.\n\nThen do the part no script can: open every page image and walk the printed checklist. One hit means a whole-document sweep for that class of issue. Fix the three most visible defects per round, re-run `--deliver`, and stop after three rounds; list what is still wrong rather than looping. If the host cannot view images, send the image paths and the checklist to the user instead of skipping the pass. A missing CJK serif shows no fallback boxes, only a heavier, flatter page, which is why the font gate exists; never call a CJK document visually verified without it.\n\nOther surfaces:\n\n- **Landing pages and any browser surface**: `--deliver` checks the HTML only. Screenshot the responsive matrix in every shipped locale before calling it done (`references/design.md` «Responsive screenshot verification»); do not wait for the user to ask about mobile.\n- **Editable PPTX**: `python3 scripts/build.py --verify slides` (or `slides-en`) and `--check-rhythm`.\n- **Missing dependencies or fonts**: `python3 scripts/build.py --doctor` reports what is installed; `bash scripts/ensure-fonts.sh` recovers CJK fonts (the commercial TsangerJinKai02 never ships inside the skill package). Render failures and their fixes: `references/production.md` Part 4.\n\n**Fresh review.** After `READY` and the image pass, review once from the contract rather than from your own rationale: reread `brief`, the coverage result, and the page images; check every acceptance item and `preserve` boundary; fix P0/P1 findings, each tied to the page or element that proves it. Use an isolated reviewer when the host has one.\n\n**Definition of done.** The closing message gives:\n\n1. The path of every deliverable, in every promised format.\n2. The `--deliver` verdict and page count against the contract.\n3. Every `[DATA NEEDED]` gap and every assumption you made, listed.\n4. The visual verdict stated honestly: when page images could not be inspected, say \"build verified, visuals unconfirmed\", not \"done\".\n\n## Feedback protocol\n\nWhen the user gives visual feedback (\"looks off\", \"太挤了\", \"not elegant\"), inspect the current render before asking them to choose a value. The render is the evidence; the user's negative label is the acceptance signal.\n\n1. Name the concrete defect in one sentence: page or surface, viewport or state, and whether it is density, hierarchy, alignment, type, color, cropping, or text fit.\n2. Lock the repair boundary: `target` is allowed to change; `preserve` names the adjacent pages, sections, content, and shared tokens that must stay stable. Ask only when two plausible targets would produce materially different artifacts.\n3. Make the smallest content, geometry, spacing, typography, crop, or token change that fixes the defect. Never hide a content problem by shrinking type first.\n4. Verify the affected matrix rather than one screenshot:\n - PDF: target page, neighboring pages, total page count, font result, and every locale or template variant reached by a shared token.\n - Screen: the breakpoint, tablet, and baseline matrix in `references/design.md` «Responsive screenshot verification»; every shipped locale; affected default, focus/selected, loading, empty/error, and transition states only when the surface has them.\n - PPTX: editable source plus a rendered PDF or opened-deck inspection.\n - Generated asset: target slot at its smallest display size plus sibling assets in the same deliverable.\n\nNever say \"I'll adjust the spacing\" without naming the exact property and its new value.\n\n**Escalate after two rounds.** If the same element is still not approved after two adjustment rounds, stop nudging values and produce one comparison artifact: the current state plus 2-3 labeled variants (A/B/C) of the same content in the product's actual frame and background, changing only the compared property. For choices with no objective criterion (typeface, accent color, logo), start there: a specimen sheet of up to 5 labeled half-page blocks with identical title-plus-paragraph content. After the pick, apply it everywhere and rebuild what it touches in the same round.\n\n## When not to use this skill\n\n- The user explicitly wants Material, Fluent, or a Tailwind default: a different design language.\n- Dark, cyberpunk, or futurist aesthetics (this is deliberately anti-future).\n- Saturated multi-color palettes (this has one accent).\n- Cartoon, animation, or illustration style (this is editorial).\n- Dynamic web app UI (this is for print and static documents).\n",".agents/plugins/marketplace.json":"{\n \"name\": \"kami\",\n \"interface\": {\n \"displayName\": \"Kami\"\n },\n \"plugins\": [\n {\n \"name\": \"kami\",\n \"source\": {\n \"source\": \"local\",\n \"path\": \"./plugins/kami\"\n },\n \"policy\": {\n \"installation\": \"AVAILABLE\",\n \"authentication\": \"ON_INSTALL\"\n },\n \"category\": \"Productivity\"\n }\n ]\n}\n","evals/equity-report-charts-en/prompt.md":"---\nplugins: [\"../../plugins/kami\"]\ntags: [equity-report, en, charts]\nmax_turns: 150\ntimeout_seconds: 2400\nallowed_tools: [Read, Glob, Grep, Skill, Bash, Write, Edit, TodoWrite]\n---\n\nWrite an equity research note as a PDF in the current directory. The company is Northgate Robotics, a made-up name for an internal training case, so use only the numbers below and add no outside data. Don't ask me questions, just finish it. Name the source file report.html and the PDF report.pdf.\n\n- Business: warehouse picking robots sold with a software subscription.\n- Revenue: FY2023 $412M, FY2024 $538M, FY2025 $701M. Show this as a bar or line chart.\n- FY2025 revenue mix: robot hardware 61%, software subscriptions 27%, services 12%. Show this as a donut chart.\n- Gross margin: FY2023 34.0%, FY2024 36.5%, FY2025 39.2%.\n- Share price $46, 180M shares outstanding.\n- Our view: FY2026 revenue $880M, rating Overweight, 12-month target price $58.\n- Risks: two customers make up 38% of revenue; a chip supply shortage could delay deliveries.\n","site/llms.txt":"# Kami\n\n> A warm parchment design system for AI-assisted professional documents. One accent color, serif fonts and clear heading levels, and consistent spacing.\n\n## Author\n- Name: Tw93\n- Site: https://tw93.fun\n- Stable entity ID: https://tw93.fun/#person\n\n\n## What is Kami\nKami gives AI agents templates and layout rules for documents and landing pages. It includes one-pagers, resumes, portfolios, slides, long documents, letters, equity reports, and changelogs. Defaults include a warm parchment background (#f5f4ed), ink-blue accents (#1B365D), and Charter/TsangerJinKai serif fonts. A white-background option is available for printing.\n\nKami installs into the agent that uses it and runs on that machine. There is no hosted Kami API, no account, and no API key: the integration surface is a local MCP server, JSON content schemas, and CLI checks.\n\n## When to use Kami\nReach for Kami when the user wants a finished document whose appearance matters, and a plain text answer would not be enough:\n- \"Make me a resume / CV\" -> `resume`\n- \"Turn this into a one-pager, product brief, or intro sheet\" -> `one-pager`\n- \"Write a formal letter, offer, or notice\" -> `letter`\n- \"Build a slide deck for this talk\" -> `slides` (PDF, or editable PPTX)\n- \"Write this research, spec, or report up as a document\" -> `long-doc`\n- \"Show my work, case studies, or portfolio\" -> `portfolio`\n- \"Analyze this company or stock and produce a report\" -> `equity-report`\n- \"Write release notes for this version\" -> `changelog`\n- \"Draw an architecture diagram, flowchart, timeline, or chart to embed\" -> one of 18 inline SVG diagram types\n- \"Build a landing page in this style\" -> browser-only `landing-page` template\n\nDo not reach for Kami for chat answers, code files, or content where the user only wants the text. Kami's cost is layout work; it pays off when the artifact is handed to someone else.\n\nHow an agent should call Kami: install the skill, fill the matching content schema, fill the template HTML, then finish with `python3 scripts/build.py --deliver filled.html content.json`, which runs every document check, renders the PDF, and exports page images for the perceptual pass, or drive the same steps as tools through the local MCP server. Judgment lives in SKILL.md; execution lives in the scripts.\n\n## Key Pages\n- [English showcase](https://kami.tw93.fun): Design system overview with live demos\n- [Developers](https://kami.tw93.fun/developers): Agent and developer integration surface\n- [About](https://kami.tw93.fun/about): What Kami is and who maintains it\n- [Contact](https://kami.tw93.fun/contact): Support and reporting channels\n- [Privacy](https://kami.tw93.fun/privacy): What this site and the skill collect\n- [Chinese showcase](https://kami.tw93.fun/index-zh.html): Chinese version\n- [Japanese showcase](https://kami.tw93.fun/index-ja.html): Japanese version (best-effort CJK path)\n- [Korean showcase](https://kami.tw93.fun/index-ko.html): Korean version (best-effort CJK path)\n- [Traditional Chinese showcase](https://kami.tw93.fun/index-tw.html): Traditional Chinese version\n- [GitHub](https://github.com/tw93/kami): Source code and templates\n\n## Install\n- Claude Code, Codex, Cursor, and other agents: `npx skills add tw93/kami -a claude-code codex cursor -g -y`\n- Claude Code plugin (v2.1.142+): `/plugin marketplace add tw93/kami` then `/plugin install kami@kami`\n- Codex plugin: `codex plugin marketplace add tw93/kami && codex plugin add kami@kami`\n- Claude Desktop: download the release asset `kami.zip` from GitHub Releases (not the source-code ZIP) and upload it via Customize > Skills > \"+\" > Create skill\n- MCP client: `claude mcp add kami -- python3 /skills/kami/scripts/mcp_server.py`\n\n## Templates\n- 8 document template types: One-pager, Letter, Resume, Long document, Portfolio, Slides, Equity report, Changelog\n- 18 diagram types: architecture, architecture board, flowchart, quadrant, bar chart, line chart, donut chart, state machine, timeline, swimlane, tree, layer stack, venn, candlestick, waterfall, sequence, class, ER\n\n## Agent Interfaces\n- Content contracts: `references/schemas/.json` carries the per-type structure and quality bar; new `content.json` files also carry an artifact `brief` with target, preserve boundary, evidence, and acceptance checks; validate before layout and check coverage after filling with `python3 scripts/build.py --check-content content.json [filled.html]`\n- Deterministic checks: placeholders, markdown residue, page density, orphan lines, slide rhythm, and a perceptual pass that exports page images with a fixed review checklist (`--check-visual`)\n- MCP server: `python3 scripts/mcp_server.py` speaks MCP over stdio with tools kami_templates / kami_doctor / kami_render / kami_check / kami_deliver / kami_screenshot, so agents can diagnose, render, and verify without loading the skill prompt\n\n## Machine-Readable Files\n- MCP server card: https://kami.tw93.fun/.well-known/mcp/server-card.json\n- Agent Skills index: https://kami.tw93.fun/.well-known/agent-skills/index.json\n- Skill definition: https://kami.tw93.fun/SKILL.md\n- Homepage as Markdown: https://kami.tw93.fun/index.md (a request to `/` carrying `Accept: text/markdown`, or `/?mode=agent`, redirects here)\n- Developer surface: https://kami.tw93.fun/developers.md and https://kami.tw93.fun/developers/llms.txt\n- Template and diagram catalog: https://kami.tw93.fun/feeds/catalog.jsonld (indexed by https://kami.tw93.fun/schemamap.xml)\n\n## Links\n- GitHub: https://github.com/tw93\n- Blog: https://tw93.fun\n- AI Profile: https://tw93.fun/llms-full.txt\n","AGENTS.md":"# Kami Agent Guide\n\n## Project\n\nKami is a document-generation skill and template system: self-contained editorial HTML\ntemplates rendered to PDF / PPTX / PNG, plus reference specs, demo assets, and a\npackaged skill archive. `SKILL.md` is the runtime manual for producing a document.\nThis file is the maintenance guide for changing the repository itself, and it records\nthe traps that a fresh read of the code does not reveal.\n\n## Layout\n\nOne repository, three top-level roles, and the install tools only ever copy the first:\n\n- `skills/kami/` is the skill and the only thing users install: `SKILL.md`,\n `CHEATSHEET.md`, `VERSION`, `LICENSE`, `references/`, `scripts/`, and the light\n `assets/` (templates, diagrams, logo, the two small font files). `npx skills add\n tw93/kami` copies exactly this directory; `bash scripts/package-skill.sh` zips it\n as `kami/` for Claude Desktop; `plugins/kami/skills/kami/` is a generated copy of it\n inside the Claude and Codex plugin. Ignored render output lives in\n `skills/kami/assets/examples/`. Every skill-relative path in `SKILL.md`\n resolves inside this directory, so nothing here may reach up into the repo.\n- `site/` is kami.tw93.fun. The Vercel project's Root Directory is `site` (project\n setting, not a file), so only this tree deploys and `site/vercel.json` is the only\n Vercel config. It carries the pages, `styles.css`, `llms.txt`, `robots.txt`,\n `sitemap.xml`, `vercel.json`, `.well-known/`, `feeds/`, and the heavy showcase and\n demo assets. Skill facts it shows (`SKILL.md`, the discovery files) are generated\n into it, never hand-copied. `/SKILL.md` and `/kami-skill.md` send\n `X-Robots-Tag: noindex`: agents can still fetch the markdown, and search\n engines should not score that file as an HTML page.\n- The root holds repository tooling only: `scripts/` (`build_metadata.py`,\n `package-skill.sh`, `release_gate.py`, `draft-release-notes.py`, `tests/`),\n `assets/fonts/` (the commercial TTFs; run `bash skills/kami/scripts/ensure-fonts.sh`\n once after cloning to copy them into the skill's ignored `assets/fonts/`, otherwise\n templates fall back to jsDelivr), `evals/` (the end-to-end skill eval suite, see\n Evals below), `docs/`, and `.github/`.\n\nRun skill-side commands from `skills/kami/` (`cd skills/kami && python3\nscripts/build.py --check`); the paths below are written from that directory unless\nthey start with `scripts/build_metadata.py`, `scripts/package-skill.sh`,\n`scripts/release_gate.py`, or `scripts/tests/`, which live at the root. Paths under\n`site/`, `plugins/`, `docs/`, `.github/`, `.claude-plugin/`, and `.agents/` are also\nrelative to the repository root.\n\n## Repository Map\n\nOnly the entries whose role is not obvious from the filename:\n\n- `SKILL.md` - skill routing plus the document-side build and verify commands.\n `CHEATSHEET.md` - quick design reference. Both ship inside the package.\n- `references/design.md`, `writing.md`, `production.md`, `diagrams.md` - full specs.\n `production.md` Part 4 is the single source of truth for render failures, their\n verified causes, and their fixes; add pitfalls there, not here.\n `docs/release.md` - release notes, release flow, demo screenshot regeneration.\n `anti-patterns.md`, `resume-writing.md`, `mermaid.md`, `deck-preflight.md`,\n `brand-profile.md` and `brand.example.md` - scoped guides.\n- `references/tokens.json` (color tokens, drift-checked by `scripts/tokens.py`),\n `references/mermaid-theme.json` (Kami to beautiful-mermaid theme, kept in sync with\n `tokens.json`), `references/checks_thresholds.json` (rhythm / density / orphan /\n visual thresholds read by `checks.py` and `visual.py`). These are live inputs to\n gates: editing a number changes what passes.\n- `references/schemas/` - one JSON Schema subset per document type. The `$comment`\n fields carry the per-field quality bar distilled from `writing.md`, so schema edits\n and `writing.md` edits move together.\n- `scripts/shared.py` - the canonical registries (`HTML_TEMPLATES`,\n `SCREEN_TEMPLATES`, `PPTX_TEMPLATES`, `MARP_TEMPLATES`, `DIAGRAM_TEMPLATES`) and each template's\n `build_max_pages`. `build.py` derives its internal build targets from them;\n Marp stays discovery-only because it uses an external CLI. Add or remove a\n template or diagram here, never in a per-script dict.\n- `scripts/render.py` - the single render entry (`render_pdf`, `build_slides`, PDF\n metadata stamping). `build.py`, `verify.py`, and `mcp_server.py` all call it; never\n open a second WeasyPrint call site.\n- `scripts/diagram_geometry.py` checks annotated architecture SVG through the existing\n lint entry points. Keep geometry in the actual rect/line/polyline attributes,\n never a second JSON source. Mark node rectangles, edge shafts and label masks;\n leave boundary frames and arrowheads unmarked. Static mask checks do not replace\n rendered text-fit and arrowhead inspection.\n- `scripts/deliver.py` - `build.py --deliver`, the one command `SKILL.md` gives an\n agent to finish a document: every document gate in a fixed order, the render, page\n images, and a single READY / NOT READY verdict. It owns the order; the individual\n `--check-*` flags stay for debugging one gate. Template family comes from content IR\n `type`, else from stylesheet overlap with the registered templates.\n `deliver(args, result)` also fills a dict with the same outcome for the MCP\n `kami_deliver` tool; `mcp_server.DELIVER_GATE_RULES` maps every deliver gate to a\n `kami_check` rule or declares it deliver-only, and `test_mcp` fails on a new gate\n that is in neither.\n- `scripts/chart.py` - `build.py --chart` / `--check-charts`: data spec to SVG for\n bar, line, hbar, donut, candlestick, and waterfall. Chart geometry lives only here;\n each SVG embeds its spec in `data-kami-spec` and the drift gate regenerates and\n compares the marks. The five chart files in `assets/diagrams/` are its output for\n their demo data: change a chart's look in `chart.py` and regenerate them, never\n edit their SVG by hand (`--check` fails on drift).\n- `scripts/brand_extract.py` - drafts a brand profile from a public website (colors,\n fonts, logo candidates, one contrast-checked accent). It fetches only the page and\n same-site stylesheets and never writes into `~/.config/kami/`.\n- `scripts/html_visibility.py` - shared static HTML/CSS visibility analysis for\n `content.py` coverage and `checks.py` residue checks. Keep its conservative\n coverage and residue modes distinct; it does not implement browser layout.\n- `scripts/tests/test_build.py` - the stable test entry point; sibling `test_*.py`\n modules group tests by responsibility, with counters and fixtures in `support.py`.\n `scripts/tests/corpus//{flag,pass}/` is the labelled corpus for the PDF\n gates (orphans, density, resume balance), rendered and judged by `test_corpus.py`.\n Before changing a value in `checks_thresholds.json`, run it and read the per-gate\n `CORPUS:` line; a fixture that flips is the cost of the change, not noise.\n- `scripts/mermaid_normalize.py` - re-themes a beautiful-mermaid SVG to the Kami\n palette and makes it WeasyPrint-safe. Pure Python, no Node, ships in the package.\n- `scripts/mcp_server.py` - zero-dependency MCP stdio server exposing\n `kami_templates` / `kami_doctor` / `kami_render` / `kami_check` /\n `kami_deliver` / `kami_screenshot`, so an MCP-capable agent can diagnose, render, and verify\n without reading `SKILL.md`. Register\n with `claude mcp add kami -- python3 /skills/kami/scripts/mcp_server.py`.\n- `scripts/math_render.py`, `scripts/mathjax_svg.js`, `scripts/ensure_mathjax.sh`,\n `scripts/mathjax-runtime/` - strict LaTeX to MathJax SVG; `render.py` imports\n `math_render`. This is the only Node dependency (Node 20 or 22+, refused otherwise),\n installed on demand by `ensure_mathjax.sh` from the tracked `package-lock.json`;\n all four ship in the package allowlist.\n- `scripts/site_facts.py` - public-site fact drift checks (install commands, version,\n template and diagram counts across `site/index*.html`, the root `README.md`,\n `site/llms.txt`), wired into `build.py --check`; it resolves those files through\n `shared.REPO_ROOT` / `shared.SITE_ROOT` and skips itself in an installed skill.\n- `scripts/check-update.sh` - quiet daily update check invoked from `SKILL.md`;\n writes a local cache marker, then resolves the latest published GitHub Release\n at most once per day. It uploads no user document or task content and stays\n silent on failure.\n- `site/assets/showcase/` - README and public-site screenshots only.\n `site/assets/demos/` - README showcase demos. Neither is part of the skill.\n- `assets/diagrams/src/*.mmd` - Mermaid source of the sequence / class / er diagrams.\n `assets/templates/marp/` - the Markdown-first Marp deck variant.\n- `dist/` is ignored. The release archive is built by `release.yml` from\n `skills/kami` and uploaded as a release asset; nothing tracks it.\n- `plugins/kami/`, `.claude-plugin/marketplace.json`,\n `.agents/plugins/marketplace.json`, and the site's discovery files are\n **generated**; see Generated Mirrors below.\n- Public site surface under `site/`: `index.html` plus `index-zh|en|ja|ko|tw.html`,\n the English-only prose pages `developers|about|contact|privacy.html`, `styles.css`,\n `llms.txt`, `robots.txt`, `sitemap.xml`, `vercel.json`. `styles.css` is Kami's own site shell\n (language switcher, gallery, responsive behavior, `.hero.doc` / `.prose` for the\n prose pages); generic template rules never belong there.\n- Agent-facing site surface, also under `site/`: `index.md` (Markdown twin of the homepage; `vercel.json`\n *redirects* `/` here for `Accept: text/markdown` and `/?mode=agent`, because Vercel\n applies `rewrites` only after the filesystem and `/` always matches `index.html`),\n `developers|about|contact|privacy.md`, `developers/llms.txt`, and the generated\n `.well-known/agent-skills/index.json`, `.well-known/mcp/server-card.json`,\n `feeds/catalog.jsonld`, `schemamap.xml`, `kami-skill.md` (served at `/SKILL.md` through a rewrite; it must not be named `SKILL.md`, or the skills CLI installs the site). Every new prose page needs its `.md` twin,\n a `rewrites` entry for the extensionless URL, and a `sitemap.xml` row.\n The `has`-conditioned redirects and the `Link` headers are only observable on a\n deploy: verify them with `curl -sI` against the preview URL, never locally.\n- `.github/workflows/check.yml` (PR/push CI) and `release.yml` (tag-triggered build\n and asset upload).\n\nReference docs are English-first and never forked per language. Inline CJK examples\nare fine where the rule itself is about CJK typography (term annotation, punctuation,\nspacing); language-specific output differences (CN/EN/KO) live in templates, not in\nduplicated reference files.\n\n## Commands\n\n`python3 scripts/build.py --help` prints the authoritative flag list and the module\nmap. Read it instead of trusting any copy; hand-maintained lists here have gone stale\nbefore. The commands it does not cover:\n\n```bash\n# from the repository root\npython3 scripts/build_metadata.py # regenerate plugins/kami, marketplaces, site discovery files\npython3 scripts/build_metadata.py --check # drift check for the same\nbash scripts/package-skill.sh dist/kami.zip # build the Claude Desktop archive from skills/kami\npython3 scripts/tests/test_build.py # full test suite\npython3 scripts/draft-release-notes.py V1.4.0..HEAD --version V1.4.1 --title \"Steadier Hand\"\nclaude plugin eval . --runs 2 --ablation none -j 4 --keep-temp --trust-plugin --no-publish \\\n --threshold 0 --judge-model sonnet --allow-tools Bash Write Edit # end-to-end skill evals\npython3 evals/score.py # deterministic gates over the newest eval run\n# from skills/kami\nbash scripts/ensure-fonts.sh # recover missing or truncated CJK fonts\npython3 scripts/mcp_server.py # MCP stdio server (render / check / screenshot)\npython3 scripts/mermaid_normalize.py raw.svg -o clean.svg\n```\n\n## Working Rules\n\n- `SKILL.md` is loaded whole on every trigger, so it carries routing, the execution\n contract, hard prohibitions, `--deliver`, and the feedback protocol, nothing else.\n Type-specific procedure (page density, resume recruiter pass, landing site mode,\n charts, illustrations) lives in the reference the tier table already sends the\n agent to. Before adding a paragraph to `SKILL.md`, name the eval case it fixes;\n before removing one, run the suite and compare against the last baseline.\n\n- Style changes must update `references/design.md` and the matching template tokens.\n- A CSS snippet in a reference doc is a shipped artifact, not prose: an agent copies\n it before it reads a template. Every fenced `css` / `html` block is scanned by\n `--check-docs` (inside `--check`) with the template rule set, and every `var()` it\n names must resolve to a registered token or one a shipped template defines. Teach\n from a component a template actually has; a recipe for assembling a new container\n is how a document ends up carrying three unrelated emphasis languages. Tag a\n deliberate counter-example inline with `/* avoid */` so the scan reads it as the\n lesson rather than the violation.\n- A change touching template tokens, shared CSS gestures, or `references/design.md`\n visual rules must rebuild the affected demo outputs (`site/assets/demos/*.pdf` / `*.png`)\n in the same change, not as a later cleanup. Demos inline their CSS by copy, so they\n silently keep the old style otherwise. Report the sweep: rebuilt N demos, K\n unaffected. The off-palette guard in `scripts/lint.py` scans `site/assets/demos/*.html`\n for stale hexes as a backstop, but it cannot see rendered PDFs or PNGs.\n Repository scans must reject an empty expected demo set; installed skills may\n skip the site because it is outside their distribution boundary. Test both modes\n with a known violating demo whenever repository paths change.\n- Templates intentionally inline their CSS rather than share a `_kami.css` partial:\n each template must stay a single self-contained HTML file the user can copy-paste\n with no build step. Fix CSS drift by applying the same change across the affected\n templates, never by introducing a build-time include.\n- For document or template tasks, lock the output contract before editing: language,\n template, output format, page or length target, visual acceptance check, and\n verification command.\n- Prefer the nearest existing template and deterministic verifier. Do not add a\n template, shared CSS layer, dependency, script flag, or optional mode unless the\n current request cannot be satisfied without it. A new template copies the nearest\n existing one, stays aligned with `references/design.md`, and adds demo coverage; a\n new document type also needs a schema in `references/schemas/`.\n- Slides default to WeasyPrint HTML-to-PDF templates unless the user explicitly needs\n editable PPTX output.\n- Mermaid diagrams: never embed raw beautiful-mermaid SVG into a PDF-bound template.\n WeasyPrint cannot resolve `color-mix()`, render ``, or fetch a\n runtime web font, so always pipe through `scripts/mermaid_normalize.py` first\n (`--check` enforces this). `xychart-beta` is browser-only because it styles through\n `