Your own app in one evening: a Mini App

8 Min. Lesezeit

No filler: take the blueprint of a working Telegram app, build your own version from it and publish it to the catalogue. You don't write code and you don't invent the idea. Time: one evening. Budget: nothing. Skills: copy, paste, and click Accept.

How this works

The old route was: come up with an idea, research the market, write a spec, explain to the model what you want, and hope. Every one of those steps is a place where it falls apart.
The order is different now. appss.pro has a ranking of Telegram apps that already work and already have an audience. Each one has a Remix button. Press it and you don't get a three-line prompt — you get a full decomposition of the app: screenshots of every screen, the logic behind every button, the navigation map, monetisation, features, target audience, and a 50-page methodology of Telegram development.
All of it comes in one archive. Download it, unpack it, hand it to cursor — it reads and builds.
The first time I opened one of those archives I worked out where the catch was: there isn't one. It's a complete teardown of an app with hundreds of thousands of users — every screen, every mechanic, every flow. A product manager spends a week assembling a document like that. Here it's one button and ten seconds.

Part 1. Picking a blueprint

You need three tools, all with a free tier: appss.pro — where you pick the blueprint and later publish the finished app; cursor — which builds the app from the blueprint; telegram — where the app lives and runs.
01. Go to appss.pro. You don't need to register just to look at the ranking.
02. Open the ranking. Five columns: by revenue, by active audience, growing, trending, new. Look at the numbers: users, growth, category. A proven mechanic beats a raw hypothesis.
03. Pick an app and hit Remix. You're not copying someone else's work — you take the skeleton and make your own version: different theme, different audience, different content.
04. Study the map: how a user moves through the bot and the mini app, what happens on each button, use cases and screenshots of every screen.
05. Download the archive with the Download button.

What's inside the archive

This isn't a file with a prompt in it — it's a complete build kit. Four parts.
PROMPT.md — the brain. 150+ lines: what the app does, how it works, features, monetisation, audience, architecture. Goes into cursor with no edits.
screens/ — the eyes. Screenshots of every screen: home, features, profile, store, settings. Cursor builds the interface from them, which is why the app comes out looking like a product rather than a prototype.
schema.json — the skeleton. A graph of every screen: what leads where, which buttons, which connections. The model reads it programmatically, so not a single button gets lost.
METHODOLOGY.pdf — the textbook. 50+ pages of Telegram development practice: bots, payments, retention, notifications. It's the same in every archive — read it once and you'll understand the architecture of any app.

Part 2. Building the app

06. Install cursor. Go to cursor.com, hit the big Download button, install it like any other program. If you're unsure: cursor is an official code editor with millions of users. It isn't a virus.
07. Load the archive and start. File → Open Folder → create an empty folder. Unpack the archive straight into it. Open the chat: Cmd+L on mac or Ctrl+L on windows.
Copy and paste: “Read PROMPT.md and METHODOLOGY.pdf in this folder. Use the screenshots in screens/ as a visual reference. Build a working app as a Telegram Mini App.”
08. Wait for the build. Cursor reads the files, looks at the screenshots, works out the logic and starts building. When it asks “Create file?” — Accept; “Run?” — Run. Expect 5–15 minutes. When it finishes, the app is built.

Part 3. Moving it into Telegram

09. Run it locally. Copy and paste: “Run the app locally and give me a link I can open in a browser.” You'll get an address like http://localhost:3000. Open it and check everything works.
10. Put it online. Telegram only opens public addresses. The simple free route is Vercel. Copy and paste: “Help me deploy this app to Vercel so it's available at a public link. Explain every step as if to a beginner.” In two or three minutes you'll have an address like https://your-app.vercel.app
11. Create a bot. In Telegram find @BotFather — the one with the blue tick. Command /newbot → name → a username ending in bot. You get a token; save it.
Attach the app: /newapp → pick the bot → name, description, 512×512 icon → paste the Vercel link into the Web App URL field.

Part 4. Launch and publish

12. Test it. Find the bot in Telegram → Start → open the mini app. It works. If it doesn't: copy the error text into cursor and write “fix it”. That's usually enough.
13. Publish to the catalogue. The circle closes: Add App → listing → name, description, icon, screenshots. The listing can be generated on appss.pro.
Free traffic: thousands of people browse the catalogue looking for new apps. You don't need ads.
Statistics: visits, opens, reactions, retention, sources. You see not just how many came in, but how they behave once inside.
The full loop: ranking → Remix → cursor → Telegram → catalogue → users → statistics → improvements.

Bonus: AI and payments

Two blocks that turn an app into a product: smart features and taking money. Both connect in the same evening.
AI. You need an API key — a password for access to the model. You register, create a key, hand it to cursor. It connects everything itself from there.
Anthropic · Claude (platform.claude.com) — a strong model for hard tasks and long texts. Google · Gemini (aistudio.google.com) — a generous free tier, good for a first launch. OpenAI · GPT (platform.openai.com) — the best known, with plenty of ready examples.
Payments. @tribute is a payment bot inside Telegram. Create a subscription or a one-off payment → get a link → embed it in the app. People pay inside Telegram and the money comes to you.

What next

That was one app. The mechanic repeats: ranking → Remix → cursor → Telegram → catalogue. The second run is faster because all the tools are already installed.
I ship a couple of apps a week and keep polishing each one until people actually use it. The point isn't the count — it's finding mechanics that work and adapting them to your own audience.