name: lynx description: Text-based browser for quick searches and readable dumps of web pages straight from the terminal.
Lynx Web Browser
lynx is a terminal-only browser. It renders HTML as text, does not execute JavaScript, and is ideal for quickly searching the web or dumping readable page content without leaving the shell.
Requirements
lynxinstalled and available in$PATH(check withlynx -version).- Network access. HTTPS works out of the box; use
-accept_all_cookiesif you want to bypass cookie prompts.
Quick Start (Interactive Mode)
lynx https://example.com
Key bindings (shown in the on-screen help bar):
- Arrow keys: Up/Down to move, Right to follow a link, Left to go back.
g: open the "Go to URL" prompt./: search within the current page.q: quit (confirm withy).
Because Lynx cannot run JavaScript, modern sites may show "browser unsupported" banners. Prefer static or "lite" endpoints when possible.
Searching the Web
Google’s default interface often blocks Lynx. Use lightweight search endpoints instead:
QUERY="hacker+news"
lynx "https://duckduckgo.com/lite/?q=${QUERY}"
Tips:
duckduckgo.com/htmlor/lite/render well in Lynx and support arrow-key navigation between results.- From within Lynx, press
g, enterhttps://duckduckgo.com/lite/?q=SEARCH+TERMS, and hit Enter. - Accept cookies when prompted (press
yorAto always accept for the domain) to avoid repeated dialogs.
Dumping Pages to Text
For a quick, non-interactive snapshot, use -dump together with -nolist:
lynx -dump -nolist https://news.ycombinator.com/ > /tmp/hn.txt
-dumpprints the rendered text to stdout.-nolisthides the numbered link list appended by default, keeping output concise.- Combine with
rg,sed, or editors to extract the info you need.
Handling Cookies & Certificates
- Accept all cookies automatically:
lynx -accept_all_cookies URL. - Ignore certificate warnings (only if you trust the site):
lynx -validate=0 URL. - For repetitive workflows, export
LYNX_COOKIE_FILEto reuse cookie storage between runs.
Debug / Logging
- Press
=inside Lynx to view document info (URL, content type, headers). - Use
lynx -traceto write request/response traces toLynx.trace(helpful when sites misbehave).
Learned Lessons
Any agent who uses this skill and uncovers new workflows, edge cases, or best practices should document them here for future reference.