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

@dk/tinman

v0.6.0

Published

Deterministic black-box testing framework for CLIs and full-screen TUIs, webrat/capybara/selenium for terminals. Drives real programs through a PTY under a Bubblewrap sandbox. Early prototype.

Downloads

547

Readme

Tinman

A deterministic black-box testing framework for CLIs and full-screen TUIs — "webrat/capybara/selenium for terminals." Tinman drives real terminal programs, including real coding agents, through an embedded PTY, and never inspects application internals.

  • Capture time may infer a mechanical test plan.
  • Replay time is absolutely deterministic: no model invocation, no network.

Sandboxed execution is the default. On Linux, Tinman launches its target inside a Bubblewrap sandbox; the operator's home, environment, and network are hidden unless explicitly granted.

Status: early prototype. Breaking changes are expected at 0.1.x.

Try it

Discover the roles, names and styles a program puts on screen:

tinman inspect opencode

Capture a live session into an editable plan, then replay it with no model and no network:

tinman record opencode
tinman test tinman.yaml

Drive Tinman from a test suite in any language over newline-delimited JSON-RPC 2.0:

tinman driver

Ask it what to run, or read the manual:

tinman
tinman man
tinman completions bash

cargo install places the binary and nothing else, so nothing lands in your manual path. Redirect tinman man into /usr/local/share/man/man1/tinman.1 and run mandb, and man tinman works from then on. Emitting the page on demand is what keeps it from drifting away from the parser that enforces it.

Requirements

Linux, and Bubblewrap 0.11.2 or newer. Tinman never executes a program outside a sandbox, so this is a hard requirement rather than an optional feature, and no flag lifts it.

The version floor is real rather than cautious: Tinman mounts your working directory through an overlay so a probed program reads your files and writes into a disposable layer, and the --overlay-src and --tmp-overlay options that make that possible arrived in Bubblewrap 0.11.0.

Debian stable and Ubuntu LTS currently ship 0.9.0, so those need backports or a newer distribution. Check what you have with bwrap --version.

macOS support waits on a sandbox backend of its own. Windows is not planned; use WSL.

The Terminal Object Model

Tinman reads the rendered screen as a tree of nested regions carrying WAI-ARIA roles, accessible names, and the presentation they were drawn with. Locators bind to that model by role and name rather than to cell coordinates, so a test survives a layout change that would break a recorded keystroke.

The full vocabulary is in the bundled skill at assets/skill/SKILL.md, which is also what a coding agent reads to learn Tinman.

Built with Shipshape

This repository uses Shipshape, a context-isolated spec-driven workflow for coding agents. Install with npx skills add dmytri/shipshape --skill '*', or the experimental open-plugin build with npx plugins add dmytri/shipshape.

License

Licensed under the BSD Zero Clause License (0BSD). See LICENSE.