pi-kagi-search

A pi extension to search the web using Kagi Search.

Installation

Install via npm:

$ pi install npm:pi-kagi-search

Or install from git:

$ pi install git:github.com/richardanaya/pi-kagi-search

Configuration

Create a JSON file at ~/.pi/kagi-search.json with your API key:

{
  "kagiSearchApiKey": "your-api-key-here"
}

Replace your-api-key-here with your actual Kagi API key.

Usage

This extension provides three powerful tools for searching and extracting web content:

1. Kagi Search (kagi_search)

Perform web searches with optional content extraction and advanced filtering.

Basic search:

Search for: "machine learning best practices"

With filters:

• Limit results: limit: 20
• Specific domains only: sites_included: ["github.com", "stackoverflow.com"]
• Exclude domains: sites_excluded: ["reddit.com"]
• File type: file_type: "pdf" (for PDFs, docs, etc.)
• Time range: time_relative: "week" (day/week/month)
• Region: search_region: "US" (ISO country codes)
• Safe search: safe_search: true

Content extraction:

• extract_count: 3 - Automatically extract full markdown from top 3 results

Example: Find recent Python tutorials from educational sites

query: "Python tutorial"
sites_included: ["edu.org", "github.com"]
time_relative: "month"
limit: 5

2. Kagi Search Media (kagi_search_media)

Search for images, videos, news articles, or podcasts.

Supported media types:

• images - Photos and graphics
• videos - YouTube, Vimeo, and other video platforms
• news - News articles
• podcasts - Podcast episodes

Example: Find recent AI news articles from major outlets

query: "artificial intelligence"
media_type: "news"
time_relative: "week"
sites_included: ["bbc.com", "nytimes.com"]
limit: 10

Example: Search for Python tutorial videos

query: "Python tutorial for beginners"
media_type: "videos"
limit: 15

All filtering options available in kagi_search are also available for media searches (domains, keywords, regions, time ranges, etc.).

3. Kagi Extract (kagi_extract)

Extract clean markdown content from up to 10 web pages simultaneously.

Converts HTML to readable markdown, removes ads and navigation, and returns just the main content.

Example: Extract content from multiple pages

urls: [
  "https://example.com/article1",
  "https://example.com/article2",
  "https://example.com/article3"
]
timeout: 30

Common Use Cases

Research: Search for information with domain filtering + extract full content

kagi_search: query + sites_included + extract_count

Media discovery: Find recent videos or podcasts on a topic

kagi_search_media: query + media_type + time_relative

Content aggregation: Extract full text from multiple sources

kagi_extract: urls (from previous search results)

Technical documentation: Search only code repositories and documentation sites

kagi_search: query + sites_included: ["github.com", "docs.python.org"]