About beavr
beavr is a small collection of single-purpose browser-only utilities built and maintained by Marcello, an independent developer based in Italy. There is no company behind it, no funding round, no team to scale. Just a developer, a static site, and 281 tools that earn their place by doing one thing well.
Why this exists
Most "free online tool" sites have the same shape: a wall of popups, three ad slots above the fold, a "download our app" modal that fires the moment you scroll, and a request to sign in for a feature that should not require an account. They process your input on their servers, log it, and often share it with advertising partners. The result is a tool that takes five seconds longer to load than the task it solves, and a privacy posture you have to read three pages of policy to decode.
beavr is the opposite. Every tool runs entirely in your browser. There is no backend for the things you actually paste in, no account, no upload, no fingerprinting, no first-party analytics. The site chrome sets preference and consent values in your browser's local storage and one language cookie, none of which is used for first-party tracking. Some tools also offer opt-in local persistence for settings, history, saved events, or text you choose to enter; those values never leave your browser and are listed in the cookie policy. The site otherwise behaves as if it were a static page from 2005 that happens to be useful. If advertising is enabled, it is served by Google AdSense only after the site's consent flow; you can read exactly what that means for your data on the privacy page.
Who builds it
Marcello is a developer who got tired of opening a different heavy web app for every five-second task: a word count, a base64 encode, a JSON pretty-print, a UUID. Each of those should be a tab you can close the moment the answer is in your clipboard. beavr is the result of treating that five-second task as the product, not as a feature buried under marketing.
The whole site is open in spirit if not in licence: the source is plain
Astro components, the data is plain TypeScript, the stylesheet is plain
CSS, and the deployment is a static dist/ directory served
by stock nginx. There is no proprietary runtime, no SSR framework, no
build step that requires a paid service. If you cloned it tomorrow, you
could run it on a $5 VPS, on GitHub Pages, on a Raspberry Pi on your desk,
or on a CDN edge. That portability is the point.
How the project is organised
The site is grouped into five categories: text and writing utilities, converters (formats, units, encodings), developer helpers, generators (UUIDs, passwords, QR codes), and educational health and fitness calculations. The first four are the daily-driver tools; the last is a small set of well-known reference calculations (BMI, VO₂ max, body fat, macro splits) presented with the caveat that none of them is medical advice. Every tool runs the same way: paste your input, get a result, copy it out. Some tools offer opt-in local history or settings, but those values stay on your device and are never uploaded.
When a tool needs a library (for example, a real QR encoder, a JWT parser, a Markdown-to-HTML converter) the dependency is small, audited, and bundled into the page itself. There is no CDN fetch on first paint, no third-party script that can change behaviour between visits, and no remote font. The first byte you download is the last byte you need to see the tool work.
Editorial standards
Every tool ships with a short editorial section explaining what it does, when it is the right tool for the job, and when a different tool — or no tool at all — would be a better choice. That last part matters: a free JSON formatter should not pretend it can replace a debugger, and a BMI calculator should not pretend it can replace a doctor. Where a tool has well-known limitations (Cooper's 1968 VO₂ formula was validated on a specific cohort; MD5 is not a security primitive) we say so on the page, not in a footnote.
Where the underlying standards are messy — XML-to-JSON mapping rules, RFC 3339 vs ISO 8601, public-suffix-list edge cases — we link to the authoritative source and reproduce the behaviour we chose. We would rather you check our work than trust it.
What this site is not
- It is not a SaaS. There is no paid tier, no upsell, and no "enterprise plan" page.
- It is not a startup. There is no roadmap, no investor letter, no growth team.
- It is not a content farm. The blog and the per-tool editorial copy are written by one person, with a clear point of view, and updated when they get out of date — not on a publishing schedule.
- It is not an AI-generated site. The tools, the editorial, and the code are written by a human. Where AI assistance is used (for example, drafting copy or testing regex variants) it is reviewed and edited before it ships.
- It is not collecting your data. See the privacy page for the short version, and the in-depth breakdown for the long one.
How to get in touch
The fastest way to report a bug, suggest a tool, or correct an editorial mistake is to use the contact page. That channel is monitored by Marcello directly; there is no support team, no ticketing system, and no chatbot in front of it. You will get a reply from a person, in English or Italian, usually within a few days.
If you found a security issue — even a small one — please disclose it responsibly through the same contact channel; the disclosure policy is summarised on the contact page.
A short history
The project started in 2026 as a personal scratchpad: a single HTML file with a word counter and a base64 encoder that Marcello could open from any browser bookmark. Over the following months the scratchpad grew into a small site, and the small site grew into a collection. Today the collection contains 281 tools, ships in English and Spanish, and runs on a static build that fits in a single Docker layer. The growth was deliberately slow: a new tool gets added only when the previous one is genuinely finished and useful, not when the SEO backlog has space.
Acknowledgements
beavr stands on the work of a long list of standards bodies, open source maintainers, and reference authors. The QR encoder uses a public domain algorithm; the JWT decoder follows RFC 7519; the Markdown converter follows CommonMark; the colour science follows the W3C and the CSS Color specifications. Where a tool's result depends on a third-party list (public suffixes, time zone data, MIME types) the version pinned in the build is documented on the tool page so you can reproduce the output.