Skip to main content
Web tools enable agents to search for information and retrieve content from the internet. Search the web for current information using Brave Search or DuckDuckGo.

​
Input Parameters

​
query
string
required
Search query string
​
count
integer
default:"5"
Number of results to return (1-10)

​
Response

Returns formatted search results with titles, URLs, and snippets.
​
results
string
Formatted search results:
Results for: {query}
1. {title}
   {url}
   {description}
2. {title}
   {url}
   {description}
...

​
Usage Example

{
  "query": "Weaver AI orchestration",
  "count": 5
}
Example Output: 
Results for: Weaver AI orchestration
1. Weaver - Ultra-Efficient AI Agent Orchestration
   https://github.com/operatoronline/weaver
   High-performance AI agent framework with tool support...

2. Getting Started with Weaver
   https://docs.weaver.ai/quickstart
   Learn how to build AI agents with Weaver...

​
Search Providers

Brave Search (recommended):
  • Requires API key configuration
  • Higher quality results with structured data
  • Rate limits apply based on plan
DuckDuckGo (fallback):
  • No API key required
  • HTML scraping-based extraction
  • May have less consistent formatting

​
Provider Selection

Priority: Brave Search > DuckDuckGo 
// Brave enabled with API key
NewWebSearchTool(WebSearchToolOptions{
  BraveEnabled: true,
  BraveAPIKey: "your-api-key",
  BraveMaxResults: 5,
})

// DuckDuckGo fallback (no key needed)
NewWebSearchTool(WebSearchToolOptions{
  DuckDuckGoEnabled: true,
  DuckDuckGoMaxResults: 5,
})

​
Error Conditions

  • query is required - Missing query parameter
  • search failed: {error} - Network error or provider failure
  • Returns nil tool if no provider is enabled or configured

​
web_fetch

Fetch a URL and extract readable content (HTML to text conversion).

​
Input Parameters

​
url
string
required
URL to fetch. Must be http:// or https://
​
maxChars
integer
default:"50000"
Maximum characters to extract (minimum: 100)

​
Response

Returns extracted content with metadata.
​
url
string
The fetched URL
​
status
integer
HTTP status code (e.g., 200, 404)
​
extractor
string
Content extraction method used:
  • text - HTML converted to plain text
  • json - JSON formatted response
  • raw - Raw content returned as-is
​
truncated
boolean
Whether content was truncated to maxChars limit
​
length
integer
Length of returned content in characters
​
text
string
Extracted content

​
Usage Examples

Fetch HTML page: 
{
  "url": "https://example.com/article"
}
Fetch JSON API: 
{
  "url": "https://api.example.com/data"
}
Limit content length: 
{
  "url": "https://example.com/long-article",
  "maxChars": 5000
}

​
Content Extraction

HTML Extraction (text):
  1. Remove <script> and <style> tags and contents
  2. Strip all HTML tags
  3. Normalize whitespace (multiple spaces → single space)
  4. Remove empty lines
  5. Return clean text content
Example: 
<!-- Input HTML -->
<html>
  <head><style>body{color:red;}</style></head>
  <body>
    <script>alert('test');</script>
    <h1>Title</h1>
    <p>Content here</p>
  </body>
</html>

<!-- Extracted Text -->
Title
Content here
JSON Extraction:
  • Parses JSON response
  • Pretty-prints with 2-space indentation
  • Returns formatted JSON string
Raw Extraction:
  • Returns content as-is for non-HTML, non-JSON types

​
LLM vs User Output

ForLLM (summary): 
Fetched {N} bytes from {url} (extractor: text, truncated: false)
ForUser (full content): 
{
  "url": "https://example.com",
  "status": 200,
  "extractor": "text",
  "truncated": false,
  "length": 1523,
  "text": "Page content..."
}

​
Error Conditions

Validation errors:
  • url is required - Missing URL parameter
  • invalid URL - Malformed URL syntax
  • only http/https URLs are allowed - Unsupported scheme (e.g., ftp://)
  • missing domain in URL - URL has no hostname
Network errors:
  • failed to create request - Request setup failed
  • request failed - Network timeout or connection error
  • stopped after 5 redirects - Too many redirects
  • failed to read response - Response body read error

​
HTTP Client Configuration

  • Timeout: 60 seconds
  • Max redirects: 5
  • Idle connections: 10
  • TLS handshake timeout: 15 seconds
  • User-Agent: Chrome 120 desktop browser

​
Security Considerations

​
URL Validation

  • Only http:// and https:// schemes allowed
  • Must have valid hostname/domain
  • URL parsing validates syntax before request

​
Content Safety

  • No JavaScript execution (HTML extraction strips <script> tags)
  • Maximum content limits prevent memory exhaustion
  • Timeout protection prevents hanging requests
  • Redirect limits prevent infinite loops

​
Privacy

  • Uses standard browser User-Agent for compatibility
  • No cookies or session persistence
  • Each request is independent

Build docs developers (and LLMs) love

Get started for freeTalk to us