The feedback loop for AI-assisted development
Coding agents write code. notiduck verifies the deployed result with real checks, then gives you a live, machine-readable share URL. Your agent can read the report, you can ship the next fix, and everyone revisits the same URL after the next check runs. Start free. Two weeks for one site included, no card required.
From generated code to verified deployment
The feedback loop for coding agents
Coding agents write code; notiduck checks what actually made it to production. Give your agent a live, machine-readable result URL so it can inspect the deployed site and revisit the same URL after the next check completes.
TTFB breakdown
We measure DNS, TCP, TLS, and server wait time from each location. You see exactly where the delay is, not just an average.
SSL, domain, and security health
Track SSL and domain expiry, modern security headers (Mozilla Observatory), WWW canonicalisation, and SEO basics. Every check produces a 100-point health score so you can see how your site is doing at a glance.
Core Web Vitals, every 4 hours
We run full Google Lighthouse audits every 4 hours. Track the Core Web Vitals that affect your search rankings and user experience. We keep your reports for a full week, each sharable via a public URL.
AI readiness monitoring
Inspect the public signals AI agents can encounter with Google Lighthouse's Agentic Browsing test. We check llms.txt, WebMCP integration, accessibility, and layout stability. Useful evidence about your site's machine-readable surface, not a prediction of what an AI model will say.
Triple-geo verification
When a site appears to go down, notiduck verifies the outage from 3 geographic locations before sending an alert. If one node detects a problem, the other two double-check: no more flaky false positives.
Anti-alert fatigue
When your site is unstable and flapping, notiduck sends one verified alert, then enforces a 2-hour cooldown on duplicate notifications. You get the data you need, then silence to fix it.
Pay-as-you-go pricing
Owned and operated by GIJA HUB MB on self-hosted bare-metal. 3 EUR/month per site, pause anytime. Bootstrapped and self-funded, with pricing designed to be sustainable rather than a temporary undercut.
Give your coding agent a live result
Every monitor generates a live, machine-readable share URL that a coding agent or other tool can read directly. It is a current view of the deployed result, not a screenshot or a one-time export.
After you or your agent ships a change, revisit the same URL after the next check completes. No screenshots, no copy paste, no guessing and no new report to wire up for every deploy.
Signals after the deploy
Verified uptime alerts
When your site is flapping between up and down, traditional monitors send you a flood of notifications. notiduck sends one verified alert, then enforces a mandatory 2-hour cooldown. You get the data you need, then silence to fix it.
Performance signals to act on
Most monitors only message you when something breaks. notiduck also alerts when your performance numbers improve. When your Lighthouse scores jump because you shipped an optimisation, the Watchduck lets the team know. Optimisation is hard work: it deserves recognition.
Built for the deploy-and-check loop
A second set of eyes after every deploy
If you run one site, you're likely running several. And though you might tell clients and investors differently, they probably use the same stack. Same boilerplate. Same plugins. Same hosting layer. When one goes sideways, they all go sideways.
A successful build is not proof that the deployed site is healthy. notiduck checks the live result for response-time drift, Lighthouse regressions, SSL problems, and the other details that are easy to miss between code and production.
For basic pings across a large fleet, self-host Uptime Kuma on a VPS. It is free and it works. Use notiduck where the deployed result matters: your agent or team reads the live report, decides what to change, and checks the same URL again after the next deploy. Uptime and predictive signals keep watching after the first fix.
Why is notiduck so affordable?
notiduck is operated by GIJA HUB MB, a registered Lithuanian company, on self-owned bare-metal infrastructure alongside our other software products. No external investors, no VC growth pressure, no markup to cover a bloated cloud bill. Just sustainable pay-as-you-go pricing: 3 EUR/month per site, pause anytime.
Pay-as-you-go. No subscriptions.
3 EUR/month per site. Pause anytime. All features included.
Need a hands-off experience? Upon request, we can configure automated balance top-ups at any amount and frequency required for your fleet.
Need higher volume, an SLA with remedy, a DPA, or annual invoicing? We work with teams directly. → Contact sales
99.9% monthly uptime SLA. If the notiduck monitoring API or dashboard is unavailable for more than ~43 minutes in a calendar month, we double all account balances. Downtime means the core service is unreachable for 5+ consecutive minutes. Excludes scheduled maintenance with advance notice. We do not exclude third-party outages. If something breaks on our end, we own it.
Frequently asked questions
How is this different from standard monitoring tools?
Standard monitors are noisy. They use hardcoded thresholds that trigger on transient network blips, leading to alert fatigue. We use a multi-location consensus engine to only alert you when the shift is real and verified.
How does notiduck prevent false positive alerts?
When a site appears to go down, we don't alert based on a single check. notiduck verifies the downtime from 3 geographic locations before sending a notification. If one node detects a problem, the other two double-check. This eliminates false alarms caused by localized network routing issues, CDN blips, or temporary DNS hiccups.
What does notiduck actually do?
We ping your site every 5 minutes from 3 locations and run a Google Lighthouse audit every 4 hours. We build a statistical baseline per location and alert you when your Core Web Vitals or response times shift significantly. Think of it as performance intelligence that actually learns your server's health.
What counts as an anomaly?
We watch for sustained shifts in your site's performance. A sudden spike in one location. Multiple locations degrading together, which usually points to the origin server. Or one region falling behind while others stay healthy, often a CDN or regional network issue. Each alert includes the actual measurements so you know exactly what changed.
How often do you run Lighthouse?
Every 4 hours on a rolling window with randomised jitter, sampling every behavioral quadrant of your site's performance. We don't just check for SEO; we use Google Lighthouse as a diagnostic sensor. If your Core Web Vitals start to degrade, it often points to a server redlining or a resource bottleneck that simple uptime checks might miss.
What is in a downtime alert?
The HTTP status code or error class, the TTFB at the time of failure, and which locations agreed on the outcome. Useful for spotting maintenance pages, config errors, and blank 5xx responses without having to open the dashboard.
What notification channels are supported?
Email notifications are free and included with every monitor. SMS alerts are available as an optional paid add-on for critical incidents.
Will notiduck spam me with notifications during an outage?
No. When an outage is confirmed, notiduck sends exactly one notification, then enforces a 2-hour cooldown on duplicate down alerts. If your server is unstable and flapping between up and down, you won't get 87 messages in 20 minutes. You get the data you need, then silence to fix it.
Does notiduck only notify me when things break?
No. notiduck also sends celebration alerts when your performance numbers improve. If you deploy an optimization and your Lighthouse score jumps, the Watchduck lets you know. Optimization is hard work: you deserve the high-five.
How far in advance do you alert for SSL and domain expiry?
We track both continuously and alert 7 days before expiry. Enough warning to renew without a late night panic.
What is the AI readiness score?
The AI readiness score comes from Google Lighthouse's Agentic Browsing category. It reports rule-based checks over observable public signals such as llms.txt, WebMCP, accessibility, and layout stability as a pass ratio (for example, 3/3). Results reflect network, browser, and page state at check time. A site without forms may show 1/2 instead of 3/3 because only applicable checks count.
Does notiduck use LLMs to verify my site?
No. notiduck uses no LLMs in its verification pipeline. It runs rule-based checks over observable public signals: HTTP, DNS, TLS, timing, Lighthouse, page structure, metadata, accessibility, and machine-readable interfaces. Results reflect network, browser, and page state at check time. It does not measure training data, rankings, mentions, or citations.
Can I add multiple sites?
Yes. Every site you add is monitored from all 3 global locations at a 5 minute cadence, billed at a flat rate. No per site limits, no tiers. If you prefer not to manage manual refills for a large fleet, we can configure custom auto-top-ups for your account upon request.
Where are you hosted?
Account and monitoring data is stored in the EU (Helsinki). Monitoring workers run from EU and Canadian locations: France (Gravelines), Finland (Helsinki), and Canada (Montreal). Workers receive only the public URLs to check and a pseudonymous identifier, never names, emails, credentials, or monitoring history, and store no user data.
Do you also monitor competitor prices?
Yes, as a separate product at price.notiduck.com. Same account, same credit pool.
I already have uptime monitoring. Why add notiduck?
Uptime monitors answer one question: is it up? They do not catch the slow creep. The TTFB that went from 200ms to 800ms over three weeks. The Lighthouse score that dropped after a dependency update. The SSL cert expiring on a Tuesday. notiduck catches all of that. Keep your basic ping monitor for the fleet. Use notiduck for the sites where performance matters.