Summarize Skill

Techniques for summarizing content from various sources.

When to Use

Use this skill when the user asks:

• •"What's this link/video about?"
• •"Summarize this URL/article"
• •"Extract text from this page"
• •"Transcribe this YouTube video"

Web Pages

Using curl + text extraction

# Fetch and extract readable text
curl -s "https://example.com" | sed 's/<[^>]*>//g' | tr -s ' \n'

Using readability-cli (if available)

readable "https://example.com" --output text

PDFs

Using pdftotext (poppler)

pdftotext input.pdf - | head -500

Using Python

import PyPDF2
reader = PyPDF2.PdfReader("input.pdf")
for page in reader.pages:
    print(page.extract_text())

YouTube (best-effort)

Using yt-dlp for subtitles

# Download auto-generated subtitles
yt-dlp --write-auto-sub --sub-lang en --skip-download -o "/tmp/%(id)s" "https://youtu.be/VIDEO_ID"
cat /tmp/VIDEO_ID.en.vtt

Without yt-dlp

Use the YouTube transcript API or fetch the page and extract from JSON data.

Summarization Approach

1. •Extract raw text from the source
2. •If text is very long (>5000 words), chunk it into sections
3. •Summarize each chunk, then create an overall summary
4. •Return a structured summary with:
   • •Title/Source
   • •Key points (3-5 bullet points)
   • •Detailed summary (if requested)

Tips

• •For very long content, summarize first, then offer to expand on specific sections
• •Always cite the source URL
• •If the content is behind a paywall or login, inform the user