メニュー
Validate factual and numeric claims in Amaroad MDX slides and flag lines missing citations. Use before release or export when slides need traceable sources and a deterministic citation audit.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | fact-citation-validator |
| description | Validate factual and numeric claims in Amaroad MDX slides and flag lines missing citations. Use before release or export when slides need traceable sources and a deterministic citation audit. |
Audit slides for unsupported factual claims and report line-numbered findings.
Automated by scripts/validate-citations.ts:
decks/<deck>/*.mdxerrorwarningAccepted evidence:
http or httpshttp or https URLsSource: or similar markers, with weaker confidence if no link existsSee references/citation-policy.md when policy detail is needed.
pnpm exec tsx .codex/skills/fact-citation-validator/scripts/validate-citations.ts --deck sample-deck
error to zeropnpm exec tsx .codex/skills/fact-citation-validator/scripts/validate-citations.ts --fail-on error
--deck <name>--format md|json--window <N>--fail-on error|warningInteractive deck design consultant that guides users through structured Q&A to plan a new Amaroad slide deck before building it. Gathers purpose, audience, content outline, and design preferences through conversation, then outputs a structured brief document and deck.config.ts parameters. Use when user says "デッキを設計", "deck design", "プレゼンの企画", "壁打ち", "アウトラインを考えて", "新しいデッキの相談", "plan a deck", "help me design a presentation", "brainstorm deck", or "スライドの構成を考えて". Key capabilities: phased Q&A workflow, design preset selection, outline pattern matching, brief document generation for deck-scaffold-from-brief.
Translates Amaroad MDX slides between Japanese and English while preserving MDX/JSX structure, code fences, URLs, and asset paths. Uses Gemini API for high-quality batch translation with structural validation. Use when user says "localize deck", "translate slides", "ja to en deck", "en to ja deck", "translate this deck to English/Japanese". Key capabilities: auto language detection, scope control (body/notes/all), glob-based file filtering, dry-run preview, placeholder protection for code and URLs, and post-localization integration with preflight and overflow checks.
Scaffolds a new Amaroad slide deck from a user brief. Generates deck.config.ts and a full set of numbered MDX slide files (cover, section, content, ending). Use when user says "new deck", "scaffold deck", "create presentation", "make a deck from this brief", or "デッキを作って". Key capabilities: configurable slide count, language (ja/en), copyright, and outline pattern selection. Outputs ready-to-fill MDX structure.
Scans Amaroad MDX slides for numeric and factual claims, then flags lines that lack verifiable citations. Produces a line-numbered report with error and warning severity levels. Use when user says "citation check", "fact check slides", "source audit", "claim validator", "check sources", or "validate references". Key capabilities: heuristic claim detection, configurable citation search radius, markdown and JSON output formats, CI-friendly --fail-on flag, single-deck or all-decks scope, and accepted evidence types (URLs, Source markers, markdown links).
Generates graphic-recording-style illustrations from slide content using Gemini image generation. Takes MDX text or slide screenshots as input, produces a hand-drawn visual note that matches the deck's theme colors, and inserts it into the target slide. Use when user says "グラレコ", "graphic recording", "グラフィックレコーディング", "グラレコ風", "visual note for this slide", or "グラレコを生成". Key capabilities: theme-aware color extraction, text-based or screenshot-based input, automatic MDX insertion, 16:9 aspect ratio output.
Edits existing images via Gemini API and updates them in Amaroad slide decks. Sends the original image with an edit prompt to apply targeted modifications such as removing objects, changing colors, or adding elements. Use when user says "edit image", "fix image", "modify image", "remove the background", or the Japanese equivalents "画像を編集", "画像を修正", "画像を直して". Key capabilities: in-place overwrite or save-as-new, visual verification before and after edit, aspect ratio and resolution control, English prompt optimization for best Gemini results.