npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@retinue/tools-browser

v0.3.0

Published

Drive a real browser when a page will not yield to a fetch: navigate, read, click and type, under an isolation story written down rather than assumed.

Readme

@retinue/tools-browser

Drive a real browser for a Retinue agent, when a page will not yield to a fetch.

npm i @retinue/tools-browser

The escalation, not the default. @retinue/tools-scrape handles most pages for the cost of one request; this costs a process, memory and seconds. Every tool description says so, and a find_tools test asserts that a query about reading a page ranks web_scrape above anything here — because a browser is the more capable-sounding tool and a model will otherwise reach for it first.

Usage

import { createBrowserToolkit, type BrowserDriver } from "@retinue/tools-browser";

// Yours: a Playwright/CDP process, or a hosted service. There is no default, deliberately.
declare const driver: BrowserDriver;

const toolkit = createBrowserToolkit({ driver, maxLifetimeMs: 5 * 60_000, maxSessions: 3 });

No browser is bundled. Chromium is ~150MB and its build has to match the driver's; a package that downloads a binary on install fails in every locked-down environment it will be deployed into. You provide the browser.

What keeps this defensible

  • References, never coordinates. Every click and keystroke names an element the model saw in a read. click(412, 890) means something different at another window size, and clicking the wrong thing succeeds. An interaction invalidates the references it came from, so a stale reference is refused rather than acted on.
  • No credential ever reaches a page. No password, token or apiKey argument exists, and browser_type refuses input[type=password] whatever the text is.
  • Hard session caps on lifetime, memory and concurrency — and teardown kills the process group, because a browser is a tree and killing the launcher leaves the renderers running.
  • The same SSRF checks as tools-scrape, from @retinue/agentkit/tools, re-run on the URL the page actually landed on so a self-redirect cannot reach internal network space.
  • Rendered text is untrusted content, fenced exactly as a scraped page is.

The isolation argument — what the platform's Sandbox port gives, why a process that needs network cannot have its strongest guarantee, what replaces it, and the one risk this does not close — is in docs/30-browser-isolation.md.

MIT.