Ideas welcome
Share an idea
ronutz is built to grow, and the clearest signal for where it should go comes from the people using it. Found a bug, a mistake, or an inaccuracy? Want a tool that isn't here yet? See a better way to handle something, or a result you'd word differently? Send it, every kind of input is welcome.
What you can send
Bugs, mistakes, and inaccuracies of any kind: a tool that misbehaves, a wrong result, an error in a Learn article, or anything that simply looks off. Feature requests for tools that already exist. Ideas for new tools the toolbox should have. Corrections and additions to the Learn articles, like a clearer explanation, a better source, or a topic that is missing. Or simply a different angle on a problem. Rough is fine; a sentence is enough to start a conversation.
If you're proposing a new tool
One test decides which floor a tool belongs on. The main floor holds tools a computer can work out just from what you type, by following a fixed, published rule, entirely in your browser. The green room takes tools that run the same way but whose form is still settling. The red room is for the rare tool whose whole job is to ask a live service, a domain registry for example: it names the official service on screen and sends nothing until you press Ask. Anything that needs to sign you in or remember you fits none of them.
Fits: decoding or explaining something you paste in (a token, a certificate, a config, command output), converting between formats, calculating from a standard or formula, or generating from a rule, like a UUID, a hash, or a command line.
Does not fit: anything that needs an account, a login, or saved data; anything that probes or scans a system that is not yours; or anything whose answer is not fixed by a published standard. A live lookup (querying a live DNS server, asking a registry who holds a domain) does not fit the main floor, but may fit the red room, under its rules.
Not sure which side your idea falls on? Send it anyway and say what it should do. I will tell you honestly whether it fits, and why.
Tools here are small, self-describing modules: a manifest that says what the tool is and where its correctness comes from, one pure function that does the work, and a set of golden vectors, the fixed input-and-output pairs that prove it. A good fit computes locally and deterministically (the same input always yields the same output, with no clock, network, or randomness in the result), keeps anything sensitive on the device, and pins its correctness to a cited source such as an RFC rather than to opinion. You don't need to build any of that to propose one: just describe what it should compute, an example, and the source it rests on.
It's worth checking the roadmap first, to see what's already planned or shipped and avoid duplicates: see the roadmap
How to reach me
Email is the channel. Tell me what you found or what you'd like, with enough detail to act on: an example, a link, the exact wording, whatever fits. If it makes the site better, it gets built.
On building new tools
In the interest of total honesty: these tools are built with CONCORD, my AI-assisted build process, and it runs on real, recurring costs. Three paid AI subscriptions carry it, one for each working seat.
- ANVIL, the engineering seat, runs on Claude from Anthropic, and this is the load-bearing one: it does the actual building, so it has to sit on one of Claude's Max tiers for there to be enough capacity to design, write, and test new tools at all.
- SCOUT, the strategy and brand seat, runs on ChatGPT Plus from OpenAI.
- PRISM, the design seat, runs on Google AI Pro.
On top of the three seats, a Cloudflare Workers plan keeps the site served, and the domains renew once a year for a fee of their own.
All in, it comes to roughly USD 150 to USD 250 a month, depending mostly on which Claude tier the engineering seat is running that month.
When the tooling is funded, tools get built. When it is not, building simply pauses until it is, which is why new tools tend to arrive in bursts rather than on a fixed schedule. You can read how the seats work, and who sits in each, on the colophon.
If you would like to help it along, buy me a coffee ☕︎. Every contribution goes into a single reserve spent on nothing but keeping that toolchain running: the three subscriptions, the hosting, and the domains. No cut, no surprises.
Right now: the tooling is funded and new tools are actively being built.
Improve the translations
English is this site's source. The Brazilian Portuguese is written with AI and reviewed gradually, and fourteen more languages are machine-translated: serviceable, and not native. If you read any of them and something is wrong, such as a term of art rendered literally or a sentence that parses and means nothing, telling me is a real contribution, and a short note naming the page and the phrase is enough.