The combination is the product.
Taken one at a time, Genkan has no unique feature. Every single thing on this list already exists somewhere else, and in several cases it exists in a more polished form.
Under the bonnetWhat "DNS-level" buys you, and exactly where it stops
Filtering at the DNS layer means the block happens when a device asks where a site is, before a single byte of it is fetched. That is why it covers an app the same as a browser, and why there is nothing to install on a phone.
It also means the honest limits are structural, not bugs. A browser doing its own encrypted DNS lookups can walk around it, which is why plain DNS is force-redirected and the common encrypted resolvers are blocked at their addresses. A device on mobile data has simply left the network. And a page served from an address shared with something allowed cannot be told apart by address alone. Every product on this page has those same three holes; the difference is which of them say so.
The claim is not that any one piece is new. It is that these seven usually come from seven different vendors, and here they are one box in your hallway.
- Whole-home parental control at the network layer. Every device on the kids' side, including the ones with no app and no account.
- Per-category active-time budgets. The clock runs while gaming is actually being used and stops when it is not, rather than a bedtime schedule or a blunt on and off.
- Network and smart-home security in the same box. A camera that may talk to its maker and to nothing else.
- Every byte stays in the house. No account, no cloud, no telemetry, nothing to leak or subpoena.
- Learn to earn. Minutes bought with quizzes and chores, graded on your own machine. This is a busy category on phones and tablets. It is empty at the network layer.
- A rewarded circumvention challenge. A household bug bounty that pays a child for beating the filter and showing their working.
- Open source. MIT, every line readable, every rule changeable.
So far we have not found a product that does all seven. That is a claim about our searching, not a proof, and it is the kind of claim that quietly rots as the market moves. If you know something that does, open an issue and this page gets edited.
Two of the seven need their claims trimmed before anybody quotes them. Learning for screen time is not a rare idea: there is a healthy market of phone and tablet apps doing it, and at least one of them launched this year. Every one of them runs on the child's device. What we could not find was any product enforcing at the network layer where a child earns their minutes back rather than being granted them by a parent, and Circle and Gryphon both have a "rewards" feature that is parent-granted bonus time. Degrading a connection instead of cutting it is not our idea either. Firewalla shipped it first, and the section on them below says so.
What this page is not. It is not a takedown of anybody. These are real products from real companies, several of them are better engineered than Genkan is, and every one of them is a reasonable thing for a parent to buy. Where a competitor is stronger, it is said plainly below rather than buried.
When you should buy something else.
Being useful to a parent means saying when the answer is not you. Four cases, and they cover most families.
You want it working tomorrow
Buy a Firewalla. If what you want is a supported appliance with a warranty, a company behind it, an app that has been through many hands, and no maintenance for you to do, Firewalla is the better product today and it is not close. It is the closest thing technically to what Genkan does at the network layer, and it is years further along.
Genkan is an old computer, a second network adapter and a few evenings. It has no warranty, no support line and one maintainer.
You want the easiest possible install
Use what your router already has, or an app-based service. Conventional parental controls install in minutes and ask nothing of you. Genkan asks you to put a machine between your router and your kids' wifi, and that is a genuine evening's work even with an AI agent doing the typing.
Circle is the name most parents know here and it is worth knowing what happened to it. The plug-in home device is gone: the product page on Circle's own shop now redirects to an app checkout for Aura, the company that bought them, and the shop's catalogue is empty. What is sold today is a phone app on a monthly plan. If a shop still lists the box, check before you buy.
If nobody in the house enjoys this sort of thing, the tool that gets set up beats the tool that is theoretically better.
Your worry is messages, not minutes
Look at Bark. If what keeps you awake is bullying, grooming or a child in trouble and not saying so, then message and content monitoring is the capability you actually want, and Bark is built for it. Genkan does not do this and will not. The section below explains why, and it is a real trade rather than an obvious win.
Note that this is Bark's app and account product, not the Bark Home box. The box does network filtering and schedules, much like everything else on this page.
You already run your own firewall
Stay on OPNsense, pfSense or OpenWrt. If you have a router platform you know and trust, you do not need another box. What Genkan adds on top is the metering, the earning and the kid-facing portal, and there is nothing stopping you lifting those ideas.
Genkan is deliberately readable for exactly this reason. The firewall rules and the schema are the documentation.
What each of these actually is.
Before the feature grid, the shape of each product. Prices move and tiers change, so treat every figure here as an indication and check the vendor before you buy.
| Product | What it is | Cost model | Where the data lives | Source you can change |
|---|---|---|---|---|
| Genkan | Software you run on a spare PC with two network ports | Free. No tier, no subscription, nothing to buy | Your house Nothing is uploaded anywhere | All of it MIT |
| Firewalla | A network security appliance with strong parental features | One-off hardware purchase, no monthly fee for the ordinary features. A paid cloud portal exists, aimed at professionals | Mixed Filtering runs on the box. The app pairs through their cloud | No Proprietary |
| Circle | The long-standing conventional choice. The plug-in home device is no longer sold: the shop is empty and the product page redirects to an app checkout for the company that bought them | Monthly subscription for the app. No hardware to buy any more | Cloud The service holds the history | No Proprietary |
| Gryphon | A family wifi router with the parental controls built into the wifi | Router purchase, with optional monthly tiers for network protection and for covering phones away from home | Cloud Cloud managed | No Proprietary |
| Bark | Two things sold together: an app and account service that monitors messages and content, and Bark Home, a network box that filters and schedules | Subscription for the monitoring. The box is a one-off purchase and there is a filtering-only plan for people who want just that | Cloud The analysis happens on their service | No Proprietary |
| OpenWrt, OPNsense, pfSense | Open router and firewall platforms. A foundation, not a product | Free software. Hardware is yours | Your house | All of it |
| AdGuard Home, Pi-hole | A DNS filter. One component, not an equivalent | Free | Your house | All of it |
Genkan runs AdGuard Home inside it. Comparing the two is a category error, and it matters because "I already have a Pi-hole" is the most common reason a parent thinks they are done. A DNS filter blocks domains. It does not meter time, it does not know which child is holding the device, and it cannot cut anything off, because a DNS server is not a firewall. Genkan uses AdGuard for the filtering and adds the firewall that makes the filtering hold.
Credit where it is due on the bypass problem. Any DNS filter can be walked past by a browser doing its own encrypted lookups, and that includes the one inside Genkan. Pi-hole handles this better than anyone else here: it ships three mitigations switched on by default, each with the reasoning written down, and its documentation says plainly that some devices have hardcoded lookups and that you should force everything through the filter at the firewall. That last sentence is the whole design of Genkan, and Pi-hole wrote it down first. AdGuard Home, which we use, does not document the problem at all in its own material, and that is a fair criticism of a thing we depend on.
Two entries deserve a sentence rather than a row. Gryphon builds the controls into the wifi itself, which is a tidy answer if you were replacing your router anyway: one box, one app, nothing extra in the hallway. It is proprietary and cloud managed, and that is the trade.
OpenWrt, OPNsense and pfSense are not products at all. They are the foundations somebody could build a Genkan on, and a household already running one of them is well placed to. What none of the three has is a time budget of any kind: they schedule by clock and calendar, and that is the end of it. OpenWrt ships thousands of packages and not one of them is a parental control. OPNsense's own plugin list marks its web proxy unmaintained. pfSense deprecated its content filter over security holes it could not get fixed and has not replaced it, and its paid edition no longer has a free tier for home use.
So "just use OPNsense" is honest advice about the firewall and misleading advice about the rest. The metering, the earning and the page a child sees when the clock runs out are the part that had to be written, and they are the part this project spent its time on.
The grid, with the losses included.
Scroll it sideways on a phone. Genkan loses several of these rows, and those rows are the reason the page is worth reading.
| Capability | Genkan | Firewalla | CircleApp only | Gryphon | BarkApp plus box | Pi-hole, AdGuard | OpenWrt, OPNsense |
|---|---|---|---|---|---|---|---|
| Whole-home filtering, every device | Yes | Yes | Was | Yes | Yes | DNS only | Build it |
| Per-category active-time budgets Time counted only while that category is in use |
Yes | Per app | Yes | Unclear | Schedules | No | No |
| Kids earn time by learning Quizzes graded, minutes credited |
Yes | No | No | No | No | No | No |
| A rewarded challenge to beat the filter | Yes | No | No | No | No | No | No |
| Smart home and IoT segmentation | Yes | Yes | No | Partly | No | No | Yes |
| Message and content monitoring Reading what a child sends and receives. This is Bark's app and account service, not the Bark Home box |
Refused | No | No | No | Yes | No | No |
| Nothing leaves the house | Yes | Partly | No | No | No | Yes | Yes |
| Open source, yours to change | Yes | No | No | No | No | Yes | Yes |
| Runs with no subscription | Yes | Yes | No | Tiered | Box only | Yes | Yes |
| Installs in under an hour | No | Yes | Yes | Yes | Yes | Usually | No |
| Supported product, warranty, someone to call | No | Yes | Yes | Yes | Yes | Community | Community |
| Mature, many households, years of use | No | Yes | Yes | Yes | Yes | Yes | Yes |
Read the bottom three rows again. Every commercial product on this grid beats Genkan on installation, on support and on maturity, and those are not small columns. They are usually the columns that decide whether a household still has working parental controls in six months.
The time-budget row needs its qualifiers said out loud, because it is the row we would most like to win. Firewalla genuinely has cumulative daily and weekly time limits that count only while an app is being used, aggregated across a person's phone, tablet and laptop. They are applied to a fixed list of about a dozen named apps rather than to whole categories, and their own users have been asking for category limits, which is why that cell says "per app" rather than "no". Circle's time limits by category are a good implementation of the same idea and Genkan is not the first to think of it. Bark Home does schedules and blocking rather than a running clock. Gryphon advertises daily limits and does not document how it measures them, so that cell says "unclear" rather than guessing in either direction.
Bark reads the messages. Genkan refuses to.
This is the most interesting difference on the page and the one most likely to matter to you, so it gets its own section rather than a tick in a box.
Bark is built to spot a child in trouble. It does that by looking at what they write and what they are sent.
The pitch is straightforward and it is a good one. A service watches a child's messages, social accounts and email for signs of bullying, self-harm, depression, predatory contact and the rest, and tells a parent when something looks wrong. It is not a transcript. By their own description a parent gets an alert and a snippet, when something is flagged, rather than the run of the conversation. The intent is that a parent finds out before the worst version of the story happens.
One distinction worth getting right, because it is easy to get wrong: Bark Home, the box, does not do this. Bark Home filters the web, blocks apps and sets schedules on your home wifi, which is roughly what everything else on this page does. The monitoring is the app and account side, on a paid tier. If you buy the box expecting the alerts, you have bought the wrong half.
The other thing to check before you rely on it is which phone your child has. Apple does not let anybody read texts and calls on iOS, so Bark's coverage there is markedly thinner than on Android, and by their own account a couple of the apps parents worry about most cannot be monitored on an iPhone at all. That is Apple's decision rather than Bark's, and it is exactly the sort of detail that only turns up after you have paid.
That is a real capability, it is one Genkan does not have, and no amount of work on Genkan would produce it. Genkan sits on the network and everything modern is encrypted. It sees that a device talked to a messaging service. It never sees a word of what was said, and the only way to change that would be to install a certificate on every device and break the encryption on purpose. We will not do that, and it is written into the project as a thing that will never be built.
Why refuse, when the upside is a child in danger
Because the cost is not nothing, and it is usually stated too quietly.
- It only works while the child does not go around it. A teenager who knows their messages are read moves the conversation somewhere unmonitored, and you have lost both the visibility and the reason they might have told you.
- Somebody else has to hold the data. Analysis of that kind runs on a service, which means a company holds signals about your child's inner life. That is a real thing to hand over and it should be a deliberate decision.
- It changes what you are to your kid. There is a line between regulating a household and reading a diary, and everybody draws it in a different place. Genkan is built for families being open with each other, which is a preference, not a proof.
None of that makes Bark wrong. For a household with a specific, live worry about a specific child, the trade may be obviously worth it, and a parent who chooses it is not making a mistake.
Can see a message that says a child is being groomed or is talking about hurting themselves, and tell you about it tonight.
Cannot, and never will. It sees that a phone talked to Snapchat. Not one word of what was said.
Covers the phone wherever it is, including on mobile data, because the monitoring is on the account and the device.
Stops at the edge of your wifi. A phone on 4G has left the network entirely.
Requires a company to process signals about your child's messages, on a subscription, off your premises.
Holds nothing anybody could ask for, because there is no service and no account. That is a consequence of where it runs, not a promise.
They are not mutually exclusive. A household worried about both minutes and messages can run a monitoring product on the phones and Genkan on the network. They solve different problems and neither one substitutes for the other. If we had to pick the one sentence a parent should take from this page, it is that the choice is between two different jobs, not between two versions of the same job.
Firewalla is the better appliance.
It is worth saying twice, because a comparison page that only flatters itself is worth nothing.
Firewalla is the product Genkan is closest to and the one it has most to learn from. It is more mature, more polished, better at working out what a device on your network actually is, and it is a real appliance rather than a project. If a friend asked what to buy this weekend and did not want a hobby, that is the recommendation.
Where Genkan differs is narrower than the marketing of either would suggest. Genkan adds the earning loop, the quiz portal, the household bug bounty, and the fact that you can read and change every rule. It also runs on hardware you already own, which for some households is the whole argument.
We have also taken an idea from them. Firewalla ships a feature called Disturb, which slows traffic to an app so that it feels like a bad network rather than a wall, with levels their documentation cheerfully names "Annoying" and "Super Annoying". The insight is a good one: a child who hits a hard block goes looking for a way around it, and a child whose feed has gone laggy usually just stops on their own. Degrading rather than cutting is on the list of things to build in Genkan and it is not built yet. It came from watching what Firewalla did, and pretending otherwise would be silly. In fairness to both sides: their version is still labelled experimental, covers a handful of apps, and needs their hardware running as your router. Ours does not exist.
They are also ahead on time limits in one respect and behind in another. Firewalla's limits are real cumulative minutes counted while an app is in use, gathered across every device a person owns, which is more than most of this market manages. They apply to a fixed list of about a dozen named apps. Genkan's clock runs per category across every service in it, which is what lets gaming stop while homework and music carry on, and which does not need a vendor to add your child's game to a list first. Which of those you want depends on whether you think in apps or in categories.
One more thing they do that we should copy: their documentation states plainly that time tracking is an estimate and can be out by around a fifth, and that a downloaded video is invisible to them. That is the right way to describe a network measurement, and it is the same limit Genkan has.
An independent review of this project put it as Genkan being an impressive working prototype rather than a finished household appliance, and Firewalla being the better appliance today. Both halves of that are fair.
How young this actually is.
The uncomfortable numbers, in one place, because they are the real reason to choose a commercial product over this one.
One household The whole user base
Genkan runs in one house, with three kids in it. Everything on this site that is marked as working, works there. Nothing here has been tested against the mess of somebody else's network, somebody else's router, or a house with different devices and different habits in it.
One maintainer No team
One parent, working with AI agents. There is no company, no support rota and nobody else who knows the code. If that person stops, the project stops, and the only insurance is that it is MIT licensed and the design notes explain why every decision went the way it did.
Days old, not years Very new
The first commit in this repository is days old. A great deal has been built in that time and none of it has been in the world long enough for the ordinary problems to show up: the slow leak, the odd device, the update that breaks on a machine nobody tested.
No release process yet No versions
No tagged releases, no packaged installer, no upgrade path other than pulling the latest code. There is a test suite with a few hundred checks and it is genuinely useful, but there is nothing between a change landing and it being what you run.
Known security findings, published Some open
This project ships its own adversarial audits, which is unusual and means anybody can read the embarrassing findings. Several serious ones have been found and fixed. Several are genuinely still open, including the command-line tool connecting to the database with more privilege than it needs.
SECURITY-STATUS.md is the current answer to "is that still a problem": what is fixed and in which commit, what is open and why, and what can never be fixed at all. Read it before you rely on this in your house.
Nothing on this list is a reason not to try it. They are reasons not to expect an appliance. If you want the thing that works and is somebody else's problem to maintain, buy the appliance. If you want the thing you can read, change and keep, and you are willing to be early, that is what is on offer here.
What nothing at the network layer can do.
These are not Genkan weaknesses relative to Firewalla or Gryphon. They are the limits of the whole approach, and they apply to every network box on this page.
Touch mobile data. A phone that turns off wifi has left the network. That needs something on the device, like Family Link or a monitoring product.
Reach anything offline. A downloaded film, an offline game, a book.
Tell you what was actually watched. Genkan sees domains, not content. It cannot separate an educational video from four hours of shorts, and does not pretend to.
Stop every VPN. Known endpoints are blocked and new ones appear constantly. There is no product that truly stops a motivated sixteen year old, and any that claims otherwise is selling something.
Make the boundary hold without an argument on the network where your children spend most of their hours, which turns out to be most of them.
Give a kid a way back in that is not pleading. That is the part none of the others do, and it is the part this project exists for.
Where these facts come from.
Genkan's own column is checked against the code in the repository, and anything half built is marked as such on the map and the roadmap rather than here.
The competitor columns are compiled from each vendor's own published material and support documentation, checked in August 2026, and from an independent review of this project against the parental-control market. We have not bought and tested every one of these products. Where a cell says "partly", it usually means the product does something in that area that is not quite the same as what Genkan does, and the fair reading is "check it yourself for your household" rather than "it fails".
Most of what is said about a competitor here was read off that vendor's own support documentation, pricing page or shop, and where a claim rests on a third party saying so it is worded as such. Two things are deliberately left open because the vendor does not answer them: how Gryphon measures a daily limit, and whether a Circle box somebody already owns keeps working. Guessing at either would be worse than the gap.
There is also an obvious conflict of interest in a project writing its own comparison page. The mitigation is that everything above is checkable: the vendor pages are public, this page's source is in the repository, and the losing rows are in the same table as the winning ones.
Feature sets and prices in this market change often, and a page like this is out of date the moment a vendor ships something. If a row here is wrong or unfair to a product, that is a bug and it will be fixed the same way any other bug is. Tell us.