LYRENTH
AgentsDocsPricingBenchmarksIndex statsAboutBlogFor site ownersStatusContact
Agents

Ready-to-use agents.

Each one does the whole job, start to finish: hand it a few web pages and get back the thing you actually wanted. Lyrenth is the part that reads, and it hands your model the article itself, with the menus, banners, scripts and markup already gone. Free, open, and they run on your machine.

The library

Pick the job, not the tool.

Every agent here reads the pages you hand it and writes the answer you asked for, with a number beside each claim pointing at the page it came from. Open one to see what it takes and what it gives back.

Together

One job, four runs.

This is a morning of homework before a meeting with a company. Each run reads its own pages and stands on its own, so you can stop after any of them. The last one hands you a file.

01

Start with what they say they do

Their about page, their manifesto and their careers page.

uvx lyrenth-agents brief https://www.mozilla.org/en-US/about/ https://www.mozilla.org/en-US/about/manifesto/ https://www.mozilla.org/en-US/careers/
What they do, who they sell to, how they describe themselves, and a list of what their own pages never answer. 3 pages, 6,664 tokens of text. See this agent
02

Find out who you are talking to

The leadership page and the list of open roles.

uvx lyrenth-agents people https://www.mozilla.org/en-US/about/leadership/ https://www.mozilla.org/en-US/careers/listings/
Names and roles as the pages state them, and every opening they list, which is where a company says what it is really building. 2 pages, 5,583 tokens of text. See this agent
03

Put their product next to the one you use

An encyclopedia article for each product.

uvx lyrenth-agents compare https://en.wikipedia.org/wiki/Firefox https://en.wikipedia.org/wiki/Google_Chrome
One table of what is actually different, with the page number in every cell and a blank wherever a page stayed silent. 2 pages, 60,000 tokens of text. See this agent
04

Leave with a file, not fifteen tabs

Their three product pages, and one sentence asking for CSV.

uvx lyrenth-agents extract -q "Write the table as CSV with a header row, one row per product, and nothing before or after the CSV." https://www.mozilla.org/en-US/firefox/new/ https://www.thunderbird.net/en-US/ https://www.mozilla.org/en-US/products/vpn/
A sheet of their products, one row each, that you save as a .csv and open in a spreadsheet. 3 pages, 5,539 tokens of text. See this agent
What the fourth run wrote
ABCDEFGH
1Product NameTypeSupported PlatformsPrivacy & TrackingPriceOpen SourceAI PolicySource
2Firefoxfast, private browserWindows 8.1 and below, macOS 10.14 and below, App Store, Google Play, desktop, mobile, tabletBlocks trackers automatically, don’t sell your personal dataAI features are optional by design[1]
3ThunderbirdappAndroid, Windows, Mac, Linuxdon’t collect personal data, sell ads in your inboxFree foreveropen sourcedon’t secretly train AI with your private conversations[2]
4Mozilla VPNvirtual private networkWindowsNo logging, tracking or sharing of network data€4.99/month, €9.99/monthopen source[3]

Saved as products.csv, this is the file your spreadsheet opens. Every cell is the wording the product page used, and the blanks are the questions those pages did not answer.

How they run

One command, one key.

Nothing to install, nothing to wire together. The agents all come from one package, lyrenth-agents, and every line of it is public.

Run it in place

One line in your terminal with the pages you care about. Each agent page shows its own command, already filled in with real URLs you can paste as they are.

Bring your own assistant

With just your Lyrenth key, an agent reads the pages and prints the finished prompt with its numbered sources, ready to paste into whichever assistant you already use. Point it at any OpenAI-compatible endpoint instead and it answers on its own.

Nothing fails quietly

A page that could not be read is listed with the reason, so you always know what the answer was built from. An answer never comes back pretending it read a page it did not.