ketch

ketch

Fast, stateless CLI for web search, scraping, code search, and library docs. Always prefer --json for structured output.

Commands

# Web search
ketch search "query"                        # titles, URLs, snippets
ketch search "query" --scrape               # + full page content
ketch search "query" --limit 10 --json

# Scrape URLs → clean markdown
ketch scrape https://example.com
ketch scrape https://a.com https://b.com    # concurrent batch

# Crawl a site
ketch crawl https://docs.example.com --depth 2
ketch crawl https://example.com/sitemap.xml --sitemap --background
ketch crawl status                          # list crawls
ketch crawl status <id>                     # check specific crawl

# Code search (OSS)
ketch code "query" --lang go                # Sourcegraph (default)
ketch code "query" --lang go -b github      # GitHub Code Search

# Library docs (Context7)
ketch docs "query"                          # auto-resolve library
ketch docs "query" --library /org/repo      # direct lookup
ketch docs --resolve "glamour"              # list matching library IDs

# Config & cache
ketch config                                # show effective config as JSON
ketch config set backend searxng
ketch cache                                 # cache stats

Key flags

Flag	Description
--json	Structured JSON output
--scrape	Fetch full content from search results
--limit, -l	Max results (default 5)
--lang	Language filter for code search
-b	Backend: brave/ddg/searxng (search), sourcegraph/github (code)
--depth	BFS crawl depth (default 3)
--background	Run crawl in background, returns crawl ID
--no-cache	Bypass page cache

Local config

Feature	Backend	Endpoint
Search	SearXNG	http://homedev:8080
Code search	Sourcegraph	https://sourcegraph.com
Docs search	Context7	—

Notes

• JS-rendered pages are detected automatically and re-fetched with headless Chrome if configured.
• Crawled pages are cached — use --no-cache to force re-fetch.

Output

Inline in chat. No file needed.