p3bot/snag/skill

Fetch web pages with snag

snag <url>                 # Headless browser by default
snag <url> <url> <url>     # Each page saved to file
snag <url> > page.md
snag -o page.md <url>
snag -d out/ <url>
snag -f md|html|text|pdf|png <url>
snag --wait-for ".sel" [--timeout N] <url>
snag --info <url>          # JSON metadata
snag --url-file urls.txt
snag --url-file -          # stdin

• PDF/PNG never go to stdout; a filename is generated when -o/-d are omitted
• Logging: --verbose and --debug (mutually exclusive)

Tabs

Tabs are sorted by URL, then title, then id (not visual left-to-right). Index 1 is first in that order. Tab features need an existing debugging browser (snag --open-browser).

snag --list-tabs
snag -t 1
snag -t 2-5
snag -t "example.com"
snag -t "https://.*\\.com"
snag --all-tabs -d out/

• -t tries index, N-M, exact URL, substring, then regex. One match → stdout; many → files
• --tab and URL arguments are mutually exclusive

Browser

snag --open-browser          # visible browser, debug port on; no fetch
snag --kill-browser          # kill debugging browsers only
snag --doctor                # environment diagnostics

Authenticated Page

Launched browsers share one persistent profile: $XDG_STATE_HOME/snag/chrome on Linux (default ~/.local/state/snag/chrome; relative or empty XDG_STATE_HOME is ignored) or ~/Library/Application Support/snag/chrome on macOS. Log in, quit, then headless snag <url> keeps the cookies. Leave the window open and same-port snag <url> attaches via CDP.

The default profile is a singleton (Chrome profile lock). Extra instances use --temp-profile or --user-data-dir PATH. Those two flags are mutually exclusive. --force-headless on another port still uses the default profile unless one of those isolation flags is set.

snag --open-browser          # Log in; quit or leave open
snag <url>                   # Same profile after quit; CDP if left open
snag -t "pattern"
snag --temp-profile <url>    # Ephemeral; headless close deletes it