Changelog

What shipped, when - product and site both, newest first. Nothing appears here before it exists; the roadmap lives on the about page and stays there until it ships.

August 27, 2026

  • Alerts can wait now. Every paid monitor carries an alert delay - how long an outage has to last before the alert actually leaves. Set it in the monitor's Alerting card: off, or up to ten minutes, with two minutes the suggestion; the API takes anything up to thirty. If the site comes back inside that window nothing is sent at all - no DOWN, and no RECOVERED apologising for it a minute later. The incident still opens the moment two regions confirm it, so your history and your status page stay honest about what happened. This one reached monitors that already existed: anything on a paid plan checking faster than every five minutes was given a two-minute delay on the day it shipped, because those were the monitors paging people for blips. Set it back to off in the same card if you liked it the old way. Free monitors alert immediately, exactly as before.
  • A region failing on its own can now speak up. Two regions have to agree before a monitor is called down, which is why a site blocked in one country, or served by one bad CDN edge, has always shown as a warning on the dashboard and nothing more. Paid monitors can now ask for a DEGRADED alert when a single region has been failing for a stretch you pick - ten minutes is the suggestion - and a REGION RECOVERED when it clears. It ships off and stays off until you turn it on, because for most sites a lone failing region is exactly the noise the two-region rule exists to swallow. It never counts as downtime: uptime figures, status pages and incident history all still mean full outages only. A monitor watched from a single region has no second opinion to compare against, so this wants at least two.

August 23, 2026

  • Status pages are live. Publish one from your dashboard and it serves at statuspage.yoping.me/your-slug: current state, uptime over 24 hours, 7 days and 30 days, and a 30-day history bar per service. The page is rendered at the edge with no JavaScript at all, so it stays up and readable on the day you most need it to. The free plan carries a "Powered by YoPingMe" link; paid plans drop it and add incident history and per-region response times, and team plans can put the page on their own domain.
  • The WordPress plugin (1.5.0) can show that page on your site: choose it under Settings, then YoPingMe, and place [yopingme_status] on any page or post for a compact strip that always links to the full page. It is opt-in twice over - no page chosen or no shortcode placed means no output and no request. Described at /wordpress-plugin.

August 21, 2026

  • Correction: the MCP page and four other places here said the MCP server had "five read-only tools" and that an assistant holding your key could "never change a monitor". A sixth tool, create_monitor, shipped on August 8 behind an opt-in write scope, and nobody updated this site for thirteen days. The pages now say six, and say which one writes. The scope was always opt-in, so no key ever gained a power its owner had not ticked - but the sentence was wrong, and a wrong sentence about what a credential can do is the worst kind to leave standing.

August 17, 2026

  • Blocking your own monitoring is easier than it sounds: /probe/allowlist covers the user agent to match on, why we publish no IP list, and what to do about Cloudflare Bot Fight Mode, which no custom rule can skip.

August 16, 2026

  • The WordPress plugin cleared review and is listed in the WordPress.org plugin directory. Install it from there, connect a free account, and your monitors show up on the wp-admin dashboard. The queue took six days.
  • Status pages: /status starts a public record of how the services you depend on actually behave, measured by our own probes instead of reprinted from theirs. Stripe is live first. Every number on the page - current state, per-region response times, uptime history - is fetched as the page loads, so a failed fetch leaves you with "checking..." rather than a number we cannot stand behind. The rest follow one at a time, hand-written, as each one is finished.

August 15, 2026

  • Pro and Team are on sale. Pro is $9 a month or $90 a year, Team is $29 or $290, each opening with a 14-day trial that asks for a card. One trial per account, ever - come back for a second and you are buying, not trying. Paddle is the merchant of record, so it runs the checkout, sends the receipt, and is the name on your card statement. Checkout itself lives on your account page, because a subscription needs an account to attach to. What we have not built: changing your own plan. Pro to Team is an email to support, and it stays on the coming-soon list until that stops being true.

August 11, 2026

  • Pricing went public: Pro and Team on the homepage, with what each costs and what each raises - monitors, check interval, regions, MCP depth. The free plan did not move, and is not going to. A refund policy at /refunds and the billing terms in /terms shipped the same day, both naming Paddle as the merchant of record. Announced today, purchasable four days later.

August 10, 2026

  • A WordPress plugin: your monitors on the WordPress dashboard, with an optional footer badge, described at /wordpress-plugin. Built and submitted to the WordPress.org plugin directory; the listing arrives when their review queue does. It leaves the coming-soon list today.

August 9, 2026

  • Two new probe regions: Oregon and London went live on August 5 and this site now says so - five locations on three continents to choose your two check regions from. The probe page had it right first; the rest of the site catches up today.
  • New tool: DNS lookup - one name asked of Cloudflare and Google at once, answers and TTLs side by side. Matching answers mean your DNS change has landed; a split shows you which resolver still holds the old record.

August 8, 2026

  • The MCP server got its first write tool: create_monitor starts watching a URL, so you can add a site by asking Claude instead of opening the dashboard. It makes HTTP monitors only, and it is refused unless the key carries the mcp:write scope - a tickbox when you create the key, off by default, so an ordinary key stays harmless if it leaks.
  • Integration recipes: wiring webhook alerts into other tools, starting with Discord (a pasteable relay Worker, with the full webhook payload contract documented) and ntfy (zero glue).
  • Telegram joined the integration recipes - a BotFather bot plus the same relay-Worker pattern, plain-text messages on purpose so markup in a cause string never drops an alert.
  • New guide: how to monitor SSL certificate expiry - why calendar reminders and renewal logs both lie, and what automatic watching has to cover.

August 6, 2026

  • The response-time chart grew up: labelled axes, a hover readout, and 24-hour, 7-day, and 30-day ranges.

August 5, 2026

  • Pick your check regions. Each monitor is now checked from two regions you choose out of Frankfurt, Virginia, and Singapore, rather than from all three. If your site is blocked in one country on purpose, drop that region and it stops raising false alarms.
  • Correction, same day: an alert has always needed a second region to agree before anyone is paged - two regions, never all three. Several pages here said "all three regions must agree", which overstated it. They now say what the product actually does.
  • The probe introduces itself: its User-Agent now links to /probe, a page for anyone who finds YoPingMe-Probe in their access logs - what it reads, how often, from where, and how to allow or block it.

August 1, 2026

  • MCP server: every account can now create an MCP API key on the Account page and connect Claude (or any MCP client) to its monitoring - five read-only tools for monitors, uptime reports, incidents, performance, and failing checks. On the free plan: the last 7 days, 50 questions a day, per-monitor reports.
  • Per-monitor alert tests: a monitor page now lists the channels it alerts through, each with a Test button, and the test alert names the site it fired for - proof the path works before an outage needs it.
  • New tool: is my site down - checks a site from outside and past its CDN, where a second, cache-busted request catches a dead origin hiding behind a cached 200.
  • A features section on this site, starting with the MCP server page.
  • About page and this changelog.
  • StatusCake and Hyperping comparisons - the top-six alternatives set is complete.
  • New guide: how to set up uptime monitoring, in about ten minutes.
  • UptimeRobot free-plan terms: a dated timeline. Their December 2024 non-commercial restriction was quietly lifted in May 2026; our own pages carried the stale claim and were corrected the same day we found out. The post is the receipt.

July 30, 2026

  • New tool: what is my IP - your public IP, country, answering edge, and HTTP/TLS versions, read in-browser and stored nowhere.
  • Two new guides: ERR_SSL_PROTOCOL_ERROR, and "my site is up for me but down for others".
  • Pingdom and Better Stack comparisons.
  • FAQ page.

July 29, 2026

  • Every page on this site now answers Accept: text/markdown with a raw-markdown version of itself, for AI agents and anyone else who prefers plain text.
  • llms.txt index and a full-article corpus at /llms-full.txt.
  • Freshping shut down in March - a migration comparison for its users.

July 27, 2026

  • First six troubleshooting guides: how to check if a site is down, 502, 503, 504, DNS_PROBE_FINISHED_NXDOMAIN, and expired SSL certificates.
  • UptimeRobot and Uptime Kuma comparisons - the honest kind, where the other side wins some rows.
  • Two free tools: the uptime calculator (what 99.9% actually allows) and the downtime cost calculator.

July 26, 2026

  • YoPingMe v1 is live. Five check types - HTTP, keyword, SSL, ping, and TCP - every 5 minutes from Frankfurt, Virginia, and Singapore, with alerts by email, webhook, or Slack only after a second region agrees.
  • One plan: free. 10 monitors, 30 days of raw history, no card.
  • This site launched alongside it.