Skip to main content

create-vlog-script

Create narration-first personal vlog scripts and slide-based recording guides from rough spoken notes and a supplied media folder. Use for reflective video essays, travel or event vlogs, and video-editing briefs; do not use for producing the final edited video unless requested.

インストールへ移動

ソース情報

リポジトリ
jykim/claude-obsidian-skills
ソースの最終更新活動
2026年9月21日 02:29
検出された SKILL.md の言語
英語
スター
50
フォーク
10

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

ファイルエクスプローラー
2 ファイル

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
create-vlog-script
description
Create narration-first personal vlog scripts and slide-based recording guides from rough spoken notes and a supplied media folder. Use for reflective video essays, travel or event vlogs, and video-editing briefs; do not use for producing the final edited video unless requested.
# Create Vlog Script Turn spoken ideas and available footage into a coherent, recordable story. ## Start with the story - Transcribe and lightly polish voice notes without erasing the speaker's point of view. - Establish a clear progression: lived memory or question, change or tension, concrete experience, reflection, and a personal resolution. - Make transitions part of the narration on the following scene's slide. Do not isolate them into separate transition slides unless the user explicitly asks for title cards. - Prefer one specific event or image over general claims. Let a personal observation lead into broader context rather than the reverse. ## Work from actual assets - Inventory the supplied media folder and visually inspect representative frames before assigning a clip to narration. Do not infer a clip's subject from its filename. - Rename original files only when the user asks, using concise, truthful, chronological names. Never label an ambiguous clip as a specific event or person without visual confirmation. - Use real supplied photos, video previews, and article captures in a recording guide wherever possible. Use a clearly labeled placeholder only when an asset has not been supplied or cannot be used yet. - Distinguish actual footage, screenshots, recreated graphics, and AI-generated concept images in both captions and narration. Never present a concept image as real data or documentary footage. - For browser-unfriendly source media, use a verified preview or still in the guide and retain the original as the final editing source. ## Create a recording-slide guide When the user wants to record narration, deliver an HTML slide guide rather than a document-only outline. - Give each slide the likely final-video composition: a full-screen background or a clear media panel, narration large enough to read aloud, and a small on-screen editing cue. - Support forward/back navigation by keyboard, click, and a visible slide counter. Include a full-screen shortcut. - Put the full spoken text—including connective phrases—on the slide. The user should not need a separate script while recording. - Place visual prompts near their corresponding narration: exact supplied asset, intended crop or hold, source article link, or an explicit placeholder. - Keep the guide presentation-oriented: do not overload it with production notes that do not affect recording. ## Research and article captures - For factual rule changes, history, or current sports information, use authoritative primary sources where possible. - When an article is used as a visual, capture the headline and relevant explanatory body section—not navigation, ads, or unrelated promotion. - Add a direct link in the slide guide and a short capture cue, such as where to pause or scroll. Confirm that the captured section supports the narration. ## Final recording review Before handoff, inspect the slide order, narration, and asset-to-claim mapping. - Verify that no asset is described as the wrong play, player, event, or outcome. - Check transitions aloud: every slide should answer why the next slide follows. - Remove duplicate ideas and move the strongest concise image or line to the ending. - State only the user-facing result and link to the recording guide.
GitHubで見る