TL;DR: Seyka is up again — on a new cloud, with no data lost. To celebrate, we've reset everyone's limits and added three days to every subscription, Early Access included. We also get to tell you what this week's post was supposed to be about: Seyka now has a public API, and her MCP server has learned to use her tools. And at the end — a promo code to share. Good news is scarce these days, so here's several at once.
We meant to write about the API. Last night had other plans: first we had to move house. Our beloved moggy is home again and not empty-handed — the gifts are right below, the API a bit further down, promise.
Since the evening of October 11, Seyka is working again: chats, projects, documents, Telegram, integrations, the developer section — all of it. Your data is safe: we restored everything to its state just minutes before the outage.
To celebrate the comeback, we have:
We're sorry about the lost day. We know for many of you it was a working one — and that you have enough bad news without ours.
Nothing, sadly, unexpected. Seyka lived in Yandex's cloud. This week Yandex's "iron heart" — that's what the company calls its data centers in its official statement — turned out to be not quite so iron three times in four days: first Sasovo, then Kaluga, and on the night of October 11, Vladimir, where Seyka had lived from the very start. At about four in the morning our availability zone stopped responding. Why Yandex? Maybe it was just bad luck: data centers are big and warm and easy to spot from afar. Or maybe it's because for years Yandex kindly helped everyone not notice what was going on — and what was going on finally noticed Yandex. We'll leave all this guessing to our heirs.
The irony is that the newest member of our team — Khariton, our DevOps bot — had finished setting up additional backups to a different cloud just minutes before and was about to run a recovery drill. The drill came early, and live-fire — a word that stopped being a metaphor this year. The good news: the backups were exactly where they should be. We stopped planning anything that can't be restored from a backup a while ago.
Yandex's status page stayed brief throughout: the platform is in emergency mode, customers are advised to move elsewhere. Many are still waiting for something to be fixed. We didn't wait. We're not convinced Yandex is where the fixing needs to happen — and, far worse than falling data centers, we're not sure we're allowed even to think, in a whisper, roughly where it does.
The data was safe from the start; Seyka's house had to be rebuilt. We set everything up from scratch at a new provider: the cluster, the databases, file storage, the CDN for static files, certificates, secrets, DNS. No "CloudBro" code word and no support queue — the clouds have plenty of new neighbours right now. A lot of people are moving these days: some from one cloud to another, some further. It took a day rather than an hour, and we chose to do it carefully. If something still isn't working the way it used to, tell us in the support chat — we'll sort it out.
Everything Seyka uses herself is now yours too — over an API, in your own scripts and services. You can now turn even your beloved little Claude into Seyka. Unlike a lot of things this year, all of it does exactly what the documentation says:
To start: sign in to Seyka, create a key on the API keys page and send it as
Authorization: Bearer sk_…. The base URL is https://proxy.seyka.ru, the interactive reference
is at proxy.seyka.ru/api-docs, and the docs with examples are in
the developer section (in Russian for now).
It's most useful if you run your own AI agent: OpenClaw, Hermes and the like. Out of the box they
search through Brave — great for the English web, hopeless for the Russian one: morphology,
regions, freshness, all of it missed. There's nothing to set up by hand: give your agent a key and
the API description at https://proxy.seyka.ru/v1/openapi.json, and ask: "Make yourself a skill
for searching Yandex through the Seyka API." It reads the description and builds the tool itself,
usually in a couple of minutes. The same goes for Wordstat, screenshots and site checks. And if
your agent speaks MCP, point it at https://mcp.seyka.ru and it gets all of it at once. You pay
only for the calls, from your Seyka balance.
One caveat: some of the Yandex services the API relies on — Wordstat and Metrica among them — are, as they put it, "unstable" right now. We hope they get fixed as quickly as money used to disappear from our Yandex Cloud account. Maybe they will even return what we left there — but we are patient: understandably, they have more important things to do right now.
Seyka's MCP server used to cover only your own data from Webmaster, Metrica,
Search Console and Analytics. Now it also has every API tool: search, Wordstat, the browser,
domain diagnostics — the same key, the same address https://mcp.seyka.ru, the same prices.
If your client caps the number of tools — like Cursor — connect a single set at its own address:
mcp.seyka.ru/search, /seo, /browser, /redirekto and more; the full list is in
the docs. And for the integrations there's a new tool that finds the
right site in Webmaster, Metrica, Search Console or Analytics by its domain, even if you have more
than fifty.
If your friends or colleagues needed a reason to try Seyka, here it is. The promo code
tialoctober gives new users Early Access through December 5 — everything in Start plus
the Google integrations, just like everyone who has been with us from the beginning. It can be
redeemed until October 31 on the pricing page, under "Redeem a promo code".
Tell your friends about Seyka — the cat is back and happy to have guests.
Peace and love.