@k2works/claude-code-booster 4.5.0 → 4.7.0

package/lib/assets/.claude/skills/creating-slides/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: creating-slides
+description: Marp Markdown スライドデッキの作成・拡張・再生成を支援。PlantUML 図表、AI 画像バリアント、漫画画像バリアント、PPTX/HTML 出力に対応。`tmp/k2works-wiki/スライド/YYYYMMDD` などの日本語技術スライドフォルダの構築・更新、コードサンプル追加、ビジュアル要約スライド追加、`スライド.images.md` / `スライド.images.ai.md` / `スライド.images.ai.manga.md` の生成、Marp HTML/PPTX デッキの再生成時に使用。
+---
+
+# Creating Slides
+
+## Core Workflow
+
+1. Read the target slide folder and the reference slide Markdown first.
+   - Check `スライド.md`, `スライド.images.md`, `スライド.images.ai.md`, and any manga variant if present.
+   - Use `rg` for headings, image references, code sample lists, and reference links.
+2. Gather source material before editing.
+   - Prefer `docs/article` for conceptual content.
+   - Use `apps` or language-specific app folders for concrete code samples.
+   - Browse only when the user requests current external references or a live website.
+3. Edit `スライド.md` as the canonical source.
+   - Keep Japanese prose concise.
+   - Keep Japanese and ASCII terms separated by a half-width space.
+   - Add code slides near related language groups.
+   - Update structure, comparison tables, and references when adding samples.
+4. Regenerate derived Markdown and outputs.
+   - Run `scripts/slide_tools.py build <slide-dir>`.
+   - For AI image variants, ensure generated Markdown references `images-ai/diagram-ai-NN.png`.
+   - For manga variants, ensure generated Markdown references `images-manga/*.png`.
+5. Verify outputs.
+   - Count PPTX slides with `unzip -Z1 '<deck>.pptx' 'ppt/slides/slide*.xml' | wc -l`.
+   - Open rendered HTML with Playwright when visual layout matters.
+   - Check representative slide screenshots, especially image-heavy and code-heavy slides.
+   - Remove temporary `render-check-*.png` files before finishing.
+
+## Commands
+
+Use the bundled script from this skill directory:
+
+```bash
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py build tmp/k2works-wiki/スライド/20260602
+```
+
+Useful subcommands:
+
+```bash
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py extract <slide-dir>
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py make-ai <slide-dir>
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py make-manga <slide-dir>
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py marp <slide-dir>
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py count <slide-dir>
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py split-summary <slide-dir> images-manga/summary-manga.png
+```
+
+## Image Variants
+
+Use the `imagegen` skill when creating new AI or manga bitmap assets.
+
+Project-bound generated images must be copied into the slide folder, normally:
+
+- `images-ai/diagram-ai-NN.png` for AI diagram replacements.
+- `images-manga/manga-*.png` for manga story panels.
+- `images-manga/summary-manga.png` and split files for tall summary comics.
+
+For tall manga summaries, do not place the full vertical image on one 16:9 slide. Split it into section images and use multiple slides so text remains readable:
+
+```bash
+python ~/.codex/skills/creating-slides/scripts/slide_tools.py split-summary <slide-dir> images-manga/summary-manga.png --parts 5
+```
+
+Then replace a single summary image slide with numbered slides that reference the split files.
+
+## Manga Markdown Pattern
+
+When converting AI Markdown to a manga variant, preserve prose and replace only diagram image references with `figure.diagram` blocks:
+
+```html
+<figure class="diagram wide">
+  <img src="images-manga/manga-01-overview.png" alt="...">
+</figure>
+```
+
+Read `references/manga-css.md` when a manga variant needs CSS or layout tuning.
+
+## Validation Notes
+
+- Marp PPTX generation needs `--allow-local-files` when local images are referenced.
+- `npx --yes @marp-team/marp-cli` is the expected renderer unless the project has a local renderer.
+- PlantUML must be available for diagram rendering; if missing, report that diagram PNG generation could not be completed.
+- If HTML keyboard navigation jumps unexpectedly during Playwright checks, inspect the DOM for `section` elements and image sizes directly.

package/lib/assets/.claude/skills/creating-slides/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Create Slides"
+  short_description: "Create Marp slide decks with visual variants"
+  default_prompt: "Use $creating-slides to create a Marp slide deck with image and manga variants."

package/lib/assets/.claude/skills/creating-slides/references/manga-css.md

@@ -0,0 +1,45 @@
+# Manga Slide CSS
+
+Use this CSS pattern in Marp decks that mix text slides and manga images:
+
+```css
+section img {
+  display: block;
+  max-width: 100%;
+  height: auto;
+}
+section .diagram {
+  margin: 10px auto 14px;
+  text-align: center;
+}
+section .diagram img {
+  width: 100%;
+  max-width: 980px;
+  max-height: 420px;
+  object-fit: contain;
+  margin: 0 auto;
+}
+section .diagram.wide img {
+  max-width: 1040px;
+  max-height: 470px;
+}
+section .diagram.compact img {
+  max-width: 760px;
+  max-height: 320px;
+}
+section.manga-summary {
+  padding: 24px 60px;
+}
+section.manga-summary h3 {
+  margin: 0 0 0.2em;
+}
+section.manga-summary img {
+  max-width: 100%;
+  max-height: 600px;
+  display: block;
+  margin: 0 auto;
+  object-fit: contain;
+}
+```
+
+For tall summary comics, prefer 4-6 cropped slides over one full-height image. Verify each crop renders at roughly 800 px or wider on a 1280 px viewport.

package/lib/assets/.claude/skills/creating-slides/scripts/slide_tools.py

@@ -0,0 +1,239 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import re
+import shutil
+import subprocess
+import sys
+from pathlib import Path
+
+
+START_RE = re.compile(r"^@(startuml|startmindmap)\b.*$", re.MULTILINE)
+END_RE = re.compile(r"^@(enduml|endmindmap)\b.*$", re.MULTILINE)
+
+
+def run(cmd: list[str], cwd: Path) -> None:
+    print("+", " ".join(cmd), f"(cwd={cwd})")
+    subprocess.run(cmd, cwd=cwd, check=True)
+
+
+def extract(slide_dir: Path) -> None:
+    src = slide_dir / "スライド.md"
+    dst = slide_dir / "スライド.images.md"
+    img_dir = slide_dir / "images"
+    if not src.exists():
+        raise SystemExit(f"missing canonical markdown: {src}")
+    img_dir.mkdir(exist_ok=True)
+    text = src.read_text(encoding="utf-8")
+    out: list[str] = []
+    cursor = 0
+    index = 0
+    fence_open_re = re.compile(r"```[a-zA-Z]*\s*\n")
+    fence_close_re = re.compile(r"\n```")
+    while True:
+        start = START_RE.search(text, cursor)
+        if not start:
+            out.append(text[cursor:])
+            break
+        end = END_RE.search(text, start.end())
+        if not end:
+            raise SystemExit(f"missing @enduml/@endmindmap after offset {start.start()}")
+        index += 1
+        block = text[start.start() : end.end()]
+        (img_dir / f"diagram-{index:02d}.puml").write_text(block + "\n", encoding="utf-8")
+        before = text[cursor : start.start()]
+        m_open = None
+        for match in fence_open_re.finditer(before):
+            m_open = match
+        m_close = fence_close_re.search(text, end.end())
+        if m_open and m_close:
+            replace_start = cursor + m_open.start()
+            replace_end = m_close.end()
+            out.append(text[cursor:replace_start])
+            out.append(f"![](images/diagram-{index:02d}.png)")
+            cursor = replace_end
+        else:
+            out.append(text[cursor : start.start()])
+            out.append(f"![](images/diagram-{index:02d}.png)")
+            cursor = end.end()
+    dst.write_text("".join(out), encoding="utf-8")
+    print(f"Extracted {index} diagrams -> {img_dir}")
+    print(f"Wrote {dst}")
+
+
+def render_plantuml(slide_dir: Path) -> None:
+    puml_files = sorted((slide_dir / "images").glob("*.puml"))
+    if puml_files:
+        run(["plantuml", "-tpng", *[str(p.relative_to(slide_dir)) for p in puml_files]], slide_dir)
+
+
+def make_ai(slide_dir: Path) -> None:
+    src = slide_dir / "スライド.images.md"
+    dst = slide_dir / "スライド.images.ai.md"
+    if not src.exists():
+        extract(slide_dir)
+    text = src.read_text(encoding="utf-8")
+    text = re.sub(r"images/diagram-(\d\d)\.png", r"images-ai/diagram-ai-\1.png", text)
+    text = text.replace(
+        "section img {\n    max-height: 420px;\n  }",
+        "section img {\n    display: block;\n    max-width: 100%;\n    max-height: 420px;\n    height: auto;\n    object-fit: contain;\n    margin: 0 auto;\n  }",
+    )
+    dst.write_text(text, encoding="utf-8")
+    print(f"Wrote {dst}")
+
+
+MANGA_REPLACEMENTS = {
+    "01": ("manga-01-overview.png", "diagram wide", "概要を描いた漫画"),
+    "02": ("manga-02-side-effect.png", "diagram", "副作用を描いた漫画"),
+    "03": ("manga-03-hard-to-test.png", "diagram", "テスト困難性を描いた漫画"),
+    "04": ("manga-04-core-shell.png", "diagram", "純粋なコアと不純なシェルを描いた漫画"),
+    "05": ("manga-05-io-plan.png", "diagram", "IO の実行計画を描いた漫画"),
+    "06": ("manga-06-execution-order.png", "diagram wide", "実行順序を描いた漫画"),
+    "07": ("manga-07-resource-safety.png", "diagram", "Resource の安全性を描いた漫画"),
+}
+
+
+def manga_css(text: str) -> str:
+    text = text.replace("font-size: 0.82em;", "font-size: 0.85em;")
+    text = text.replace("font-size: 0.68em;", "font-size: 0.7em;")
+    needle = "section img {\n    max-height: 420px;\n  }"
+    repl = """section img {
+    display: block;
+    max-width: 100%;
+    height: auto;
+  }
+  section .diagram {
+    margin: 10px auto 14px;
+    text-align: center;
+  }
+  section .diagram img {
+    width: 100%;
+    max-width: 980px;
+    max-height: 420px;
+    object-fit: contain;
+    margin: 0 auto;
+  }
+  section .diagram.wide img {
+    max-width: 1040px;
+    max-height: 470px;
+  }
+  section .diagram.compact img {
+    max-width: 760px;
+    max-height: 320px;
+  }
+  section .columns .diagram {
+    margin-top: 0;
+  }
+  section .columns .diagram img {
+    max-height: 330px;
+  }"""
+    return text.replace(needle, repl)
+
+
+def make_manga(slide_dir: Path) -> None:
+    src = slide_dir / "スライド.images.ai.md"
+    dst = slide_dir / "スライド.images.ai.manga.md"
+    if not src.exists():
+        make_ai(slide_dir)
+    text = manga_css(src.read_text(encoding="utf-8"))
+    for num, (filename, klass, alt) in MANGA_REPLACEMENTS.items():
+        image = slide_dir / "images-manga" / filename
+        if not image.exists():
+            continue
+        block = f'<figure class="{klass}">\n  <img src="images-manga/{filename}" alt="{alt}">\n</figure>'
+        text = text.replace(f"![](images-ai/diagram-ai-{num}.png)", block)
+    dst.write_text(text, encoding="utf-8")
+    print(f"Wrote {dst}")
+
+
+def marp(slide_dir: Path) -> None:
+    targets = [
+        ("スライド.images.md", "スライド.html", ["--html"]),
+        ("スライド.images.md", "スライド.pptx", ["--pptx", "--allow-local-files"]),
+        ("スライド.images.ai.md", "スライド.images.ai.html", ["--html"]),
+        ("スライド.images.ai.md", "スライド.images.ai.pptx", ["--pptx", "--allow-local-files"]),
+        ("スライド.images.ai.manga.md", "スライド.images.ai.manga.html", ["--html"]),
+        ("スライド.images.ai.manga.md", "スライド.images.ai.manga.pptx", ["--pptx", "--allow-local-files"]),
+    ]
+    for src, dst, flags in targets:
+        if (slide_dir / src).exists():
+            run(["npx", "--yes", "@marp-team/marp-cli", src, "-o", dst, *flags], slide_dir)
+
+
+def count(slide_dir: Path) -> None:
+    for pptx in sorted(slide_dir.glob("*.pptx")):
+        proc = subprocess.run(
+            ["unzip", "-Z1", str(pptx), "ppt/slides/slide*.xml"],
+            check=True,
+            text=True,
+            capture_output=True,
+        )
+        print(f"{pptx.name}: {len(proc.stdout.splitlines())}")
+
+
+def split_summary(slide_dir: Path, image_rel: str, parts: int) -> None:
+    src = slide_dir / image_rel
+    if not src.exists():
+        raise SystemExit(f"missing image: {src}")
+    try:
+        proc = subprocess.run(
+            ["sips", "-g", "pixelWidth", "-g", "pixelHeight", str(src)],
+            check=True,
+            text=True,
+            capture_output=True,
+        )
+    except FileNotFoundError as exc:
+        raise SystemExit("split-summary requires macOS sips") from exc
+    nums = [int(x) for x in re.findall(r"pixel(?:Width|Height):\s+(\d+)", proc.stdout)]
+    if len(nums) < 2:
+        raise SystemExit(f"could not read image size from sips output: {proc.stdout}")
+    width, height = nums[0], nums[1]
+    stem = src.stem
+    chunk = height // parts
+    for idx in range(parts):
+        offset = idx * chunk if idx else 1
+        crop_h = chunk if idx < parts - 1 else height - idx * chunk - 1
+        out = src.with_name(f"{stem}-{idx + 1:02d}.png")
+        shutil.copy2(src, out)
+        run(["sips", "-c", str(crop_h), str(width), "--cropOffset", str(offset), "0", str(out)], slide_dir)
+        print(f"Wrote {out.relative_to(slide_dir)}")
+
+
+def build(slide_dir: Path) -> None:
+    extract(slide_dir)
+    render_plantuml(slide_dir)
+    make_ai(slide_dir)
+    make_manga(slide_dir)
+    marp(slide_dir)
+    count(slide_dir)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("command", choices=["extract", "make-ai", "make-manga", "marp", "count", "build", "split-summary"])
+    parser.add_argument("slide_dir")
+    parser.add_argument("image", nargs="?")
+    parser.add_argument("--parts", type=int, default=5)
+    args = parser.parse_args()
+    slide_dir = Path(args.slide_dir).expanduser().resolve()
+    if args.command == "extract":
+        extract(slide_dir)
+    elif args.command == "make-ai":
+        make_ai(slide_dir)
+    elif args.command == "make-manga":
+        make_manga(slide_dir)
+    elif args.command == "marp":
+        marp(slide_dir)
+    elif args.command == "count":
+        count(slide_dir)
+    elif args.command == "build":
+        build(slide_dir)
+    elif args.command == "split-summary":
+        if not args.image:
+            raise SystemExit("split-summary requires an image path relative to slide-dir")
+        split_summary(slide_dir, args.image, args.parts)
+
+
+if __name__ == "__main__":
+    main()