fetch

Fetch content from a specific URL and store it

Fetch Skill

Fetch and extract content from a specific URL. Use when you have a direct link rather than needing to search.

Arguments

The URL to fetch: $ARGUMENTS

How to Execute

  1. Extract the URL from arguments
  2. Call the fetch endpoint:
curl -X POST http://localhost:3200/api/fetch \
  -H "Content-Type: application/json" \
  -d '{"url": "<url>"}'
  1. Present the extracted content to the user:
    • Page title
    • Main content (cleaned/extracted)
    • Metadata (author, date if available)

Content Extraction

The service uses Jina Reader for intelligent content extraction:

  • Removes navigation, ads, footers
  • Extracts main article content
  • Preserves code blocks and formatting
  • Handles various page types (docs, blogs, repos)

When to Use

  • User provides a specific URL to analyze
  • Following up on a search result
  • Fetching documentation from a known source
  • Reading GitHub READMEs or wiki pages

When NOT to Use

  • Searching for information (use /research instead)
  • Need to explore multiple sources
  • Don't have a specific URL