summarize

name	summarize
description	Extracts and summarizes web content from URLs. Use when users share links or ask to summarize articles, YouTube videos, or web pages.

Summarize Skill

Extract readable content from URLs — articles, YouTube videos, podcasts, and web pages.

When to Use

Use this skill when users:

• Share a URL and want a summary or key points
• Ask to "read", "summarize", or "extract" a web page
• Paste a YouTube link and want transcript/content
• Want to understand an article without visiting it

How to Summarize

Use the telclaude summarize CLI command via Bash:

# Basic usage — extract content from URL
telclaude summarize "https://example.com/article"

# Limit output size (default: 8000 characters)
telclaude summarize "https://example.com/article" --max-chars 4000

# Get markdown-formatted output
telclaude summarize "https://example.com/article" --format markdown

# Custom timeout (default: 30000ms)
telclaude summarize "https://example.com/article" --timeout 60000

Output Format

The command outputs structured metadata followed by the extracted content:

Title: Article Title Here
Site: example.com
Words: 1234
Transcript: youtube-captions     (only for video/audio content)
Note: Content was truncated      (only if content exceeded max-chars)
---
[Extracted text content here]

Supported Content Types

• Articles/blog posts: HTML extraction via readability
• YouTube videos: Automatic caption/transcript extraction
• Podcasts: Audio transcript when available
• General web pages: Best-effort content extraction

Options

Flag	Description	Default
--max-chars <n>	Maximum characters to extract	8000
--format <fmt>	Output format: text or markdown	text
--timeout <ms>	Request timeout in milliseconds	30000

Rate Limits

• 30 requests per hour per user
• 100 requests per day per user

Tips

• For long articles, use --max-chars 16000 to get more content
• Use --format markdown when the user wants formatted output with headings/links
• YouTube URLs automatically extract captions — no special flags needed
• If extraction fails, the URL may require JavaScript rendering (not currently supported)