write-metadata

name	write-metadata
model	haiku
description	Generate YouTube and Facebook metadata for book video
disable-model-invocation	false
argument-hint	<book-slug>

Write Video Metadata

Generate YouTube and Facebook metadata for a Bookie book video.

Context

• When to use: Video rendered (or about to render) — need title/description/tags for upload.
• Input: chunks-display.md (content, shorts candidates) + storyboard.md (scene structure) + notes.md (book info, angle)
• Output: metadata.md (YouTube + Facebook metadata)
• Reference: See books/atomic-habits/metadata.md for the standard format.

Steps

1. Validate: Check that $ARGUMENTS is provided. If missing, ask Hai for the book slug (e.g., "atomic-habits"). Set SLUG=$ARGUMENTS. Check that projects/ai-book-video/books/$SLUG/ exists.

2. Read input files:

   • Read projects/ai-book-video/books/$SLUG/chunks-display.md — extract sections, [SHORT] candidates
   • Read projects/ai-book-video/books/$SLUG/storyboard.md — extract scene structure, shorts markers
   • Read projects/ai-book-video/books/$SLUG/notes.md — extract book title, author, angle, key points
   • If either file is empty or just a template scaffold, abort and tell user which file to prepare first.

3. Read reference: Read projects/ai-book-video/books/atomic-habits/metadata.md to match the exact output format.

4. Generate YouTube metadata:

   Title (3 options):

   • Vietnamese, max 60 characters
   • All Vietnamese text MUST use proper diacritics (có dấu). Exception: hashtags/tags (no diacritics OK)
   • Curiosity-driven, clear benefit
   • No cheap clickbait
   • Book title must appear in the title

   Description:

   • Book/author credit line
   • 2-3 hook sentences (from HOOK section in chunks-display.md)
   • Scene labels from storyboard.md
   • Buy links (placeholder)
   • Social links: bookiecommunity.com | facebook.com/bookie.community
   • Hashtags at end of description

   Tags (15-20):

   • Mix: book title (VN + EN), author name, topic, "tom tat sach", "review sach", "doc sach", "self improvement", keywords VN + EN, "bookie"

5. Generate Shorts metadata:

   • For each [SHORT] candidate in chunks-display.md:
     • Short title (max 40 characters)
     • 5 hashtags
     • Full short description with hook text (30-60s standalone content)

6. Generate Facebook metadata:

   • Post caption: emoji formatting, numbered key points from chunks-display.md, CTA comment
   • Hashtags (Vietnamese + English mix, include #Bookie #BookieInspiresEveryone)

7. Decision point: Present the full draft (steps 4-6 produce a draft, NOT written to disk yet). Present 3 title options. Ask Hai to pick preferred title (or suggest revision).

8. Write output: After Hai approves, write complete metadata to projects/ai-book-video/books/$SLUG/metadata.md

Important

• This skill runs after script is written. Prerequisites: chunks-display.md, storyboard.md, and notes.md with chosen angle.
• After metadata is done, next step: upload to YouTube/Facebook (manual).

Format

Output must follow this structure (matching atomic-habits reference):

# Video Metadata: [Book Title] — [Angle/Subtitle]

## YouTube
### Title (pick one)
### Description
### Tags (N)
### Shorts
### Shorts Descriptions

## Facebook
### Post Caption
### Hashtags