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

@emilia-protocol/crash-test

v0.2.0

Published

The EMILIA crash test: one command shows a high-risk action blocked, dual-approved by two named humans on their own devices, issued as an authorization receipt, then verified OFFLINE with EMILIA gone — and a forged copy rejected. Produces an audit-grade W

Readme

@emilia-protocol/crash-test

One command. Two acts. The whole protocol becomes obvious.

npx -y @emilia-protocol/crash-test

Act 1 — Authorization (at the county, today)

An AI finance agent proposes a $2,400,000 grant disbursement to a new vendor bank account. Self-approval is rejected. The Finance Director approves on her device; the Controller approves on his. A two-person quorum holds. An authorization receipt is issued.

Act 2 — Reliance (the auditor's desk, six months later)

The network is down. The EMILIA service is deleted. The database is gone. The auditor has one file. It still verifies — offline, against no one's server — and a forged copy is rejected. The auditor gets a workpaper.

Act 2 is the product. Act 1 is the setup.

Scenarios — same engine, different high-risk action

The default tells the county-finance story. Two more show the same EP-QUORUM-v1 predicate gating high-risk actions in healthcare:

npx -y @emilia-protocol/crash-test --scenario clinical

A high-alert IV medication (heparin infusion). The agent proposes administration; an independent double-check by a second qualified clinician is required (the ISMP / Joint Commission control); the order is signed, then a forged copy with the infusion rate altered 10× is rejected. The receipt carries only hashed patient and encounter identifiers — no PHI — so the evidence can be shared without a HIPAA disclosure.

npx -y @emilia-protocol/crash-test --scenario procurement

A hospital capital purchase (a $1.85M 3T MRI from a new, off-contract vendor). Dual control — Department Director, then CFO — and a forged copy with the vendor bank account swapped after approval (textbook payment-redirect / BEC fraud) is rejected, because the payee account is inside the signed action.

What you get

The run writes an Auditor Workpaper Package to ./emilia-workpaper/:

  • authorization-receipt.json — the evidence the auditor keeps.
  • verification-report.md — an audit-grade determination with a single bolded verdict: PRESENT AND INDEPENDENTLY VERIFIED or ABSENT / UNVERIFIABLE — DO NOT RELY — the absence made visible, not a silent gap.

Verify a receipt yourself (the auditor's path)

npx -y @emilia-protocol/crash-test verify ./emilia-workpaper/authorization-receipt.json

Exit code 0 = verified, 1 = do not rely. No network, no account, no API key. The check recomputes the action hash from the action as filed, runs the real EP-QUORUM-v1 predicate (verifyQuorum from @emilia-protocol/verify), and checks the operator's commit signature.

What it proves, and does not prove

Proves: the named approvers, holding their own device keys, each signed this exact action under the stated policy, in order, before execution — and no party, including EMILIA, could forge or alter it undetected.

Does not prove: absence of collusion or coercion among distinct approvers; that the displayed action matched intent (presentation integrity); the real-world identity behind each enrolled approver. Stated, not claimed solved.

Honesty note

The approver signatures here are real ES256 device-class (Class A) WebAuthn assertions, minted locally so the demo runs without hardware. In production they originate on each approver's own device. The crash test makes no network calls — that is the entire point of Act 2.

Spec: draft-schrock-ep-quorum · draft-schrock-ep-authorization-receipts

Apache-2.0