# add-video

> Download a YouTube video transcript and fully process it into the knowledge base

- Author: Ryan Kwong
- Repository: rhkwong/high-powered-engineer
- Version: 20260122172418
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-06
- Source: https://github.com/rhkwong/high-powered-engineer
- Web: https://mule.run/skillshub/@@rhkwong/high-powered-engineer~add-video:20260122172418

---

---
name: add-video
description: Download a YouTube video transcript and fully process it into the knowledge base
argument-hint: <youtube-url> [output-name]
allowed-tools: Read, Write, Edit, Glob, Grep, Bash, TodoWrite
---

# Add Video Skill

Download a YouTube video transcript and process it through the full content pipeline.

## Usage

```
/add-video <youtube-url> [output-name]
```

**Examples:**
```
/add-video https://www.youtube.com/watch?v=abc123
/add-video https://youtu.be/xyz789 meta-staff-engineer-interview
```

## Pipeline Overview

This skill orchestrates five steps:

| Step | Skill | What it does |
|------|-------|--------------|
| 1 | `/yt-script-downloader` | Downloads transcript from YouTube |
| 2 | `/generate-summary` | Creates detailed summary |
| 3 | `/update-theme-index` | Adds video to theme index |
| 4 | `/update-people-index` | Adds guest to people index |
| 5 | `/reindex` | Updates vector search index |

## Instructions

### Step 1: Download Transcript

Execute the `/yt-script-downloader` skill workflow:

1. **Validate URL**: Ensure it's a valid YouTube URL (youtube.com or youtu.be)

2. **Get video title** for default filename:
   ```bash
   yt-dlp --get-title "URL"
   ```

3. **Download transcript**:
   ```bash
   yt-dlp --write-auto-sub --write-sub --sub-lang en --skip-download --sub-format vtt -o "/tmp/yt-transcript" "URL"
   ```

4. **Convert VTT to clean text**:
   ```bash
   sed -e '/^WEBVTT/d' -e '/^Kind:/d' -e '/^Language:/d' -e '/^$/d' -e '/^[0-9][0-9]:[0-9][0-9]/d' -e 's/<[^>]*>//g' /tmp/yt-transcript.en.vtt | awk '!seen[$0]++' > /tmp/yt-transcript.txt
   ```

5. **Generate filename**:
   - If output-name provided, use it
   - Otherwise, convert video title to kebab-case

6. **Save raw transcript** to `video-transcripts/raw-transcripts/<filename>.md`:
   ```markdown
   # [Video Title]

   **Source:** [YouTube URL]
   **Downloaded:** [Today's Date]

   ---

   [Transcript content]
   ```

7. **Clean up**:
   ```bash
   rm -f /tmp/yt-transcript.en.vtt /tmp/yt-transcript.txt
   ```

### Step 2: Generate Summary

Execute the `/generate-summary` skill workflow:

1. Read the raw transcript just created
2. Create summary at `video-transcripts/summaries/<filename>.md`
3. Follow full template:
   - Quick reference (TL;DR, level focus, companies, themes)
   - Guest background & career timeline
   - Detailed insights (5-15 topics)
   - Stories & anecdotes table
   - Quotable moments
   - Contrarian claims
   - Connections (themes and people to update)

### Step 3: Update Theme Index

Execute the `/update-theme-index` skill workflow:

1. Read the summary
2. Read `video-transcripts/content-index/_theme-index.md`
3. For each theme tag in summary:
   - Add video to Videos list
   - Update consensus/conflicting views
   - Add best stories
   - Refine synthesized takeaways

### Step 4: Update People Index

Execute the `/update-people-index` skill workflow:

1. Read the summary
2. Read `video-transcripts/content-index/_people-index.md`
3. Create/update guest entry:
   - Profile, career path, insights, stories, quote
4. Add notable mentions
5. Update company and level indexes

### Step 5: Update Vector Search Index

Execute the `/reindex` skill workflow to add the new content to semantic search:

```bash
cd /Users/ryankwong/high-powered-engineer && python3 vector-search/indexer.py
```

This incrementally adds the new summary's topic chunks to the vector index so they appear in `/search` results.

## Error Handling

| Error | Action |
|-------|--------|
| yt-dlp not installed | Tell user: `brew install yt-dlp` |
| No subtitles available | Ask if user wants to provide manual transcript |
| Invalid URL | Ask for valid YouTube URL |
| Video already exists | Ask to overwrite or use different name |

## Quality Checklist

Before completing, verify:
- [ ] Raw transcript saved with metadata
- [ ] Summary captures all major topics
- [ ] Guest background complete
- [ ] At least 3 stories extracted
- [ ] Theme index updated for all relevant themes
- [ ] People index has complete guest entry

## Output

Report completion:
```
✓ Downloaded: video-transcripts/raw-transcripts/<filename>.md
✓ Summary: video-transcripts/summaries/<filename>.md
✓ Themes updated: #tag1, #tag2, #tag3
✓ People updated: [Guest Name]
✓ Search index updated: N new chunks indexed
```