Kagi LLM skill

---
name: search
description: |
  web search, crawling, and page interaction via Kagi search cli -> Firecrawl CLI . Use this skill whenever the user wants to search the web, find articles, research a topic, look something up online, scrape a webpage, grab content from a URL, extract data from a website, crawl documentation, download a site, or interact with pages that need clicks or logins. Also use when they say "fetch this page", "pull the content from", "get the page at https://", or reference scraping external websites. This provides real-time web search with full page content extraction and interact capabilities.
allowed-tools:
  - Bash(firecrawl *)
  - Bash(kagi *)
---

# search

- what? `skill to search and process the web`
- when? (
   user.ask[questions examine proof problem verify]: `user has a question`
   user.know[understand comprehend grasp fathom perceive]: `user wants to understand something`
   user.do[crawling scraping maping downloading searching]: `user wants to interact with website(s)`
  )
- where? `through system shell using cli tools[kagi firecrawl]`

## search.basic

1. kagi.search[`kagi is a better google`]: `kagi "foobar"`
- kagi.search.multiple[`multiple searches at once`]: `kagi "foobar" && kagi "fooobar"`
- kagi.search.output[`example output`]: ```
TITLE1
https://example1.com
SUMMARY1

TITLE2
https://example2.com
SUMMARY2

TITLE2
https://example2.com
SUMMARY2

--- Related Searches ---
examples of examples
list of examples
examples news```
2. redundancy.check[`skip scraping if success`]: `ls .firecrawl | grep -E 'foo|bar'`
3. firecrawl.scrape[`scrape results`]: `firecrawl scrape -u "example.com" -o .firecrawl/example.md`
- firecrawl.scrape.flag.o[`naming convention`]: ```
.firecrawl/search-{query}.json
.firecrawl/search-{query}-scraped.json
.firecrawl/{site}-{path}.md```
- firecrawl.scrape.flag.f[`output format`]: ```
# format options: markdown, html, rawHtml, links, images, screenshot, summary, changeTracking, json, attributes, branding 

# no format flag defaults to markdown
firecrawl scrape -u "example.com" -o .firecrawl/example.md
# use format to get content relevant to your task
firecrawl scrape -u "example.com" -f "links" -o .firecrawl/example.md
# multiple formats will return json
firecrawl scrape -u "example.com" -f "links,images" -o .firecrawl/example.json```
- firecrawl.scrape.multiple[`multiple scrapes at once`]: ```bash
firecrawl scrape -u "example.com" -o .firecrawl/1.md &
firecrawl scrape -u "example.com/foo" -o .firecrawl/2.md &
firecrawl scrape -u "bar.example.com" -o .firecrawl/3.md &```
4. redundancy.check[`helps reduce token usage`]: ```bash
# List word count
wc -l .firecrawl/example.md

# With head for quick peek
wc -l .firecrawl/file.md && head -50 .firecrawl/file.md```
5. os.read[`read results`]: ```bash
# Grep keyword
grep -n "keyword" .firecrawl/file.md

# Use jq to parse json
jq -r '.data.web[].url' .firecrawl/example.json

# example for getting titles and urls
jq -r '.data.web[] | "\(.title): \(.url)"' .firecrawl/example.json```

## firecrawl.commands

- firecrawl.scrape[`have a url, page is static or js-rendered`]: `get a pages content`
- firecrawl.download[`only use if kagi returns nothing`]: `Find pages on a topic`
- firecrawl.crawl[`bulk extract a site section`]: `Need many pages (e.g., all /docs/)`
- firecrawl.map[`need to locate a specific subpage`]: `find urls within a site`
- firecrawl.search[`only use if kagi returns nothing`]: `Find pages on a topic`
- firecrawl.agent[`need structured data from complex sites`]: `AI-powered data extraction`
- firecrawl.interact[`content requires clicks, scrolls, or typing, use with interact for best results`]: `interact with a page`
- firecrawl.download[`download a site to files`]: `save an entire site as local files`

#!/usr/bin/env bash

# Kagi Search CLI
# Usage: tool <search query>
# Add your API key below

API_KEY="APIKEY_HERE"

if [ -z "$1" ]; then
    echo "Usage: $0 <search query>"
    exit 1
fi

QUERY="$*"

RESPONSE=$(curl -s -G \
    --data-urlencode "q=$QUERY" \
    -H "Authorization: Bot $API_KEY" \
    "https://kagi.com/api/v0/search")

# Check for API errors
if echo "$RESPONSE" | jq -e '.error' &> /dev/null; then
    echo "API Error:"
    echo "$RESPONSE" | jq -r '.error'
    exit 1
fi

# Print search results (t=0)
echo "$RESPONSE" | jq -r '.data[] | select(.t == 0) | "\n\(.title)\n\(.url)\n\(.snippet // "")"'

# Show related searches if any (t=1)
RELATED=$(echo "$RESPONSE" | jq -r '.data[] | select(.t == 1) | .list[]' 2>/dev/null)
if [ -n "$RELATED" ]; then
    echo -e "\n--- Related Searches ---"
    echo "$RELATED"
fi