Barkme Is Live

On September 4, 2026 we put Barkme live. It watches the sites, web applications and APIs you are responsible for, from the outside, and barks the moment one of them goes down. It also keeps an eye on the things around them that fail quietly: SSL certificates, domain registrations, and the reputation of the domains you send mail from. Alerts go to Email, Slack, Telegram, Linear, Trello or your own webhook, and every project gets a public status page. You can start on the Free plan at barkme.dev.

That's the announcement. The rest of this article is about why it exists and what is inside.

Where it came from

The main function of software, in general, is to take a routine task and turn it into a process. Sometimes the routine task belongs to a client. Sometimes it is your own, and you only notice it when it has been quietly eating your time for a year.

We run our own sites: welldigit.com and the product sites. We also run web applications and APIs for clients, on their servers, in their stacks, behind their credentials. Every one of them needs the same thing: someone has to know, quickly, when it stops responding.

For a long time we solved that separately in each context. A cron job on one server. A health-check endpoint wired to a client's own Slack on another. A third client with nothing at all, because access to their infrastructure was the whole point of the engagement, and adding our monitoring there meant one more thing to maintain and hand over.

Each solution was reasonable on its own. Together they were tech debt. Different credentials, different tools, different people who knew where the alerts went. And none of them watched the site from where the users are: outside.

UptimeRobot, Pingdom and Datadog exist, and they are good at what they do. We considered them. We went our own way for two reasons. We wanted something as simple as possible, a monitor in three clicks. And we wanted alerts to arrive where we and our clients already live: Telegram, Email, Slack, Linear, Trello. A dashboard that someone has to remember to open is where alerts go to die.

So we built it, ran it on our own assets first, then on our clients', and now it is public.

What exactly needs watching

Does the endpoint return 2xx? Does it answer within a few seconds? That is where everyone starts, and it is the least interesting part, because most teams eventually notice a full outage. The expensive failures are the quiet ones.

Your domain lands on a blacklist for a reason you never learn, and security-filtering DNS resolvers stop answering for it. The server is fine. Your monitoring is green. Some of your visitors simply get nothing.

The domain you use for transactional email gets flagged as spam. Order confirmations, password resets and invoices start going nowhere, and the first person to tell you is a customer who has been waiting three days.

Or the one I am fairly sure you have met at least once: the domain expires, or the SSL certificate does. Nobody ignored the registrar's email. It was pointed at the person who set up the domain two years ago, or at a mailbox that no longer exists. The site turns into a parking page, and nobody on the team finds out until a client does.

These are attention failures more than engineering failures, and they happen in the gaps between the tools each team already has. That is the pain Barkme is built for: know quickly when anything is wrong, and know it before a client or a customer tells you.

What it does

Uptime is the entry point. The rest is what you actually need around it.

Uptime checks. Barkme sends an HTTP GET to each URL from the outside, the way a user would, every 30 seconds, minute, 2 minutes or 5 minutes. A 2xx response counts as up. Everything else counts as down: a 3xx redirect, a 4xx or 5xx, a timeout, a DNS or TLS failure. Each monitor has its own timeout, from 1 to 120 seconds, so "does it answer" and "does it answer fast enough" are the same check.

SSL and domain expiry. Every verified domain is checked once a day, and you get a warning 30, 14, 7, and 1 day before the certificate or the registration expires. Each threshold fires once. Nobody needs a daily reminder; they need one they cannot miss.

Domain reputation. Once a day, on two independent tracks. Web reputation compares your domain's DNS answer across security-filtering resolvers (Cloudflare, AdGuard, Quad9's threat feeds) against clean resolvers, so you find out the moment filters start blocking visitors. Mail blacklist looks the domain up on Spamhaus DBL, SURBL and URIBL, the listings that decide whether your email lands. One alert when the domain appears on a list, naming the sources, and one when it clears.

Public status pages. One live page per project, showing 90 days of uptime per monitor, refreshed about every minute. On the Free plan it lives on a barkme.dev subdomain. Paid plans serve it from your own domain, like status.acme.com. Monitor URLs are never shown, and the page is marked noindex.

Teams. An owner, admins and members, with the roles you would expect: only the owner touches billing, admins set up shared channels and invite people, members use the dashboard.

How it works, in plain words

Three steps.

1. Prove the domain is yours. Add one TXT record on the apex: host @, value barkme-verify= followed by the token Barkme shows you. It usually takes about two minutes, and verifying the apex covers every subdomain and every URL beneath it. You do this once per domain.
2. Add a monitor. A URL and an interval. That is the whole form.
3. Connect a channel. Email, Slack, Telegram, Linear, Trello or your own webhook. Barkme sends a real test alert through it before anything goes live, so the first bark you receive is one you asked for.

From then on Barkme knocks on your URL on schedule. When a check fails, it retries two seconds later, and it takes two consecutive failed checks before an incident opens: four failed requests in total. A single blip will not wake you. A 30-second monitor alerts in about a minute.

When the site comes back, the incident closes on the first successful check, and you get a recovery notice with the total downtime. Slow to open, quick to close.

What's inside

A few details that matter once you rely on the thing.

An incident is backdated to the first failed check, so the confirmation window counts against your uptime. That minute is part of the number on your status page.

A monitor only ever has one open incident at a time. A two-hour outage produces one alert and one recovery notice, however many checks failed in between.

Linear and Trello go further than a notification. An incident opens an issue or a card automatically and closes it when the site recovers, so nobody triages an alert that resolved itself an hour ago. Linear and Trello handle incidents only; expiry and reputation alerts go to Email, Slack, Telegram and the webhook.

The webhook posts every incident, expiry alert and reputation alert to your HTTPS endpoint as JSON, signed with an X-Barkme-Signature header: an HMAC-SHA256 of the timestamp and the raw body. Recompute it on your side and you know the request came from us.

Timestamps are stored in UTC and rendered in the timezone you pick, so switching zones never rewrites history and daylight saving cannot corrupt an incident duration.

If you ever downgrade and end up over the limits, the extra monitors are paused. Nothing is deleted. Your oldest monitors keep running, and the rest come back when you upgrade again.

These are the decisions you make after you have been on the receiving end of bad alerts.

Channels

Email, Slack and Telegram for people. Linear and Trello for the backlog. A signed webhook for everything else.

Email and Telegram are personal channels: each team member connects their own. Slack, Linear, Trello and the webhook are shared: one destination per team, set up by an owner or admin. Any channel can be muted without being deleted.

Plans

The Free plan is a real plan, not a trial. One domain, three monitors, checks every minute, expiry and reputation alerts included. Payments on the paid plans are handled by Paddle, and switching or cancelling applies straight away.

How to start

Go to barkme.dev and create an account, no card needed. Verify a domain, add a URL, connect a channel. If you want to see the whole flow first, how it works walks through it screen by screen, and the FAQ has the exact limits and edge cases.

We use Barkme for our own sites and for our clients'. It is the supervision layer that sits outside your server and outside your domain settings, and sometimes that other angle is the one that catches the problem.