Password security guide

Security and Privacy Methodology

Secret values are handled inside a separate self-hosted origin that has no advertising, analytics, storage or password endpoint.

Architecture and data-flow boundary

The main site and secure tool use different origins. Browser same-origin enforcement prevents the parent page from reading the frame DOM or JavaScript state. The parent accepts messages only from the configured secure origin and frame window, then allowlists message types.

Threat model

The design reduces exposure to content-site analytics, advertising code, ordinary page scripts, server logs and accidental URL submission. It assumes a modern browser correctly enforces origin isolation and Web Crypto. It does not defend a compromised device, malicious browser extension, hostile operating system, screen capture, clipboard monitor, phishing page or unauthorized change to the website.

Encrypted transport, server security and controlled administrative access therefore remain essential parts of the service.

Cryptographic random generation

Generators use crypto.getRandomValues(). Character selection uses rejection sampling so out-of-range random values are discarded instead of folded with biased modulo arithmetic. Group requirements are enforced by rejecting an otherwise random candidate that lacks a selected group.

Passphrase words are selected independently from the self-hosted 7,776-entry EFF long list. Strength analysis uses a self-hosted estimator. No runtime code or word list is fetched from a CDN.

What the main site can and cannot access

DataMain site accessSecure origin behavior
Generated password or passphraseNoHeld in frame memory until replaced or navigation
Strength-check input and resultNoAnalyzed locally
Frame heightYesSent for layout
Allowlisted action nameYesMay be sent without value, score or length

Browser and security-header requirements

A current browser must support Web Crypto, sandboxed frames, origin-aware postMessage and the clipboard API for copy actions. The secure origin sets a restrictive Content Security Policy, blocks framing by origins other than the canonical main site, disables unneeded permissions, blocks indexing and serves self-hosted assets only.

The main site has its own CSP and can permit consent-gated providers without adding them to the secure origin. That separation is intentional.

Responsible disclosure and verification

Report a reproducible security issue through the contact method on the contact page. Do not include real passwords, authentication cookies, private keys or personal data. State the affected URL, browser, expected behavior, observed behavior and minimal reproduction steps.

Reports are checked against the current public service. Relevant controls include cryptographic random selection, origin checks, Content Security Policy, the absence of secure-tool storage and telemetry, and correct HTTP behavior.

Known limits

No website can promise absolute safety. A strong generated password is only one control. Users must verify the correct domain, maintain device security, protect recovery channels, avoid reuse and enable modern authentication where available. The operator must protect server access, encrypted transport, backups and monitoring.

Sources and further reading

Generate a value inside the isolated toolThe main page never receives the generated password.Open password generator →
Written and reviewed by Gabor Kohanyi

Technical claims are checked against the cited primary standards and official service documentation. Corrections are handled under the editorial policy.