01 Whoami

sshlg@me ~ % whoami

Sergey Sheleg

Tech founder · AI product builder · co-founder of Nicegram · author of The Proof of Done Manifesto

I'm a tech founder and I build products. I co-founded Nicegram, a privacy-first Telegram client that reached 56M+ organic installs without paid acquisition, and I'm building Prowl, PrivateClawd and CheckMyData now. 13 years of doing that is packaged into 11 installable skill packs. They form the workflow layer of the ssheleg harness, alongside Project Observatory and the developing Asset Foundry. I wrote the standard underneath it, Proof of Done.

Mobile products
Agent systems
Growth & distribution
Go-to-market
AI cost engineering
Monetization
Market research

Work with me Install the toolkit Follow on X contact@sshlg.me

Years: 13+ shipping
Shipped: 10 products
Live: 6 running
Toolkit: 39 skills, MIT

02 Toolkit

Skills for agent work

I package the workflows I use into skills an agent can follow: deciding what to build, implementing it, checking the result and leaving evidence. The skills catalog explains each pack; the install command below adds the family.

Proof of Done: The Agentic Software Development Manifesto makes one claim: when agents write the code, done is not the last message of the run, it is a state of the system that can be proven. It asks any finished piece of work for four fields.

DONE: What became true.
PROOF: What was executed or observed, and where the result lives.
SCOPE: The commit, environment, requirements, and surfaces covered.
NOT VERIFIED: What was not checked, could not be checked, or remains uncertain.

NOT VERIFIED: none within the stated scope is a valid answer. Silence is not. The 39 skills below, across 11 packs, are the manifesto's reference implementation and the workflow layer of the ssheleg harness. I use them on the products further down and on this page.

Packs: 11
Skills: 39
Installs: 4,763
npm downloads: 6,723

Installs are what the directory recorded when somebody ran the installer. Downloads count tarball fetches of the launcher over 30 days, which includes CI runners, registry mirrors, and this machine every time it updates the family. Neither number measures active users. Both are re-read from their own source by the build.

install $ npx sshlg-skills install

One command, every agent · MIT, no telemetry · github · npm · skills.sh

6 slash commands · Claude Code plugins · Cursor, Codex, OpenCode, Gemini CLI, Windsurf, Zed and the rest of the 70+ agents the skills CLI supports

The ssheleg harness

The harness is how I connect agent workflows with tools that observe projects and produce assets. Skills supply the working methods. The projects below add capabilities that are installed and configured separately. Explore the harness ↗

Project Observatory ↗

Project Observatory brings the original observation engine into open source: project discovery, Git activity, metrics, credential tracking and exposure checks, with a local dashboard and MCP tools for agents. Each user keeps a private workspace with their own projects, configuration and keys. Provider integrations, model calls and scheduling are enabled explicitly.

The agent-guided setup starts locally, then connects the sources you choose. The existing portable commands remain available alongside the full engine. Read the source ↗

Asset Foundry · in development

I'm building Asset Foundry for agents that need to produce 3D assets, images and audio as part of a project. It is a developing component of the harness, with a separate setup from the skills.

03 Projects

What's running.

Five lead the page: two I co-founded at scale, then three I am building now. Every card carries the domain, the role I held and figures read from the product's own site, so checking one takes about a minute. The complete ledger follows, including the ones sold and the one still in early access.

Co-founded, at scale

Nicegram LIVE
nicegram.me
56M+ organic installs

A privacy-first Telegram client, and the largest thing I have co-founded.

Five years from an info-product hypothesis to a client people find and install on their own.

Growth through SEO, referrals, viral mechanics and product work. No paid acquisition.

GPL-2.0 across all three platforms, so the privacy claim is one you can read.

Co-founder · growth

TelegramGPL-2.0Open sourceGrowth iOS Android Desktop
eSIM Plus LIVE
esimplus.me
Two halves of one app: travel data over eSIM, and a second phone line.

On the App Store since December 2019 and still shipping: the current version went out in July 2026.

Regional and global plans let one eSIM cross a continent without a swap.

Virtual numbers are a real second line for SMS, calls and service registration.

Co-founder · product

eSIMiOSAndroidWebPayments App Store

Building now

Prowl LIVE
prowl.chat
448 market-intelligence tools behind one MCP endpoint.

448 tools across 17 providers, covering SEO, ads, SERP, reviews, pricing and trends.

33 modules in the catalogue, with up to 16 combined in one report.

A deep report costs about $8.60 in credits, and the free tier starts with $5 of them.

Founder · build · GTM

MCPTypeScriptNext.jsLLM orchestrationStripe
PrivateClawd LIVE
privateclawd.com
An AI agent on a dedicated VM, running in 60 seconds.

Roughly three hours of manual setup reduced to a 60-second deploy.

Token caps, loop detection and auto-pause stop an unattended agent from running without a limit.

From $30 a month for one agent, with prompt-injection and link-preview exfiltration defence included.

Co-founder · build · GTM

Cloud orchestrationDockerOpenClawStripeEdge
CheckMyData LIVE
checkmydata.ai
Ask a database a question in English and get back the SQL that answered it.

PostgreSQL, MySQL, ClickHouse and MongoDB. Read-only by default.

It indexes the codebase as well as the schema, so business logic stays part of the answer.

MIT, self-hosted with Docker, no telemetry. Data stays on the machine where it runs.

Founder · build · GTM

TypeScriptPostgresDockerLLM agentsOpen source

Two more tracks

The whole portfolio

10 shipped · 6 still running · 3 sold · since 2013

10 products I have shipped or co-founded, newest first, with the role I held on each one. Employment inside somebody else's company stays in the biography below rather than in this ledger, because a ledger of ownership is what it says it is. The Ultimate Guitar years were the largest audience I have ever worked on and they still do not belong in one.

04 Path

How I got here.

13 years of shipping, plus the years before that. The line through all of it is the same: engineering and economics are one product problem seen from two sides, and anyone who sees only one of the two builds the other one badly.

Before 2013 Mogilev · Minsk

Olympiads, then a degree that named the job.

I started programming as a child in Mogilev and won school informatics olympiads. I graduated from BSUIR as an engineer-programmer-economist, a title that sounded bureaucratic and turned out to describe the work exactly: the code and the pricing are the same problem seen twice.
2013 → 2015 Elinext · HiQo Solutions

Other people's products, other people's deadlines.

Elinext from 2013, HiQo Solutions from 2014. I shipped software to order for European and American clients, then built for companies in tech, agro, AI and hardware. This is where writing software and delivering it became two different skills, which is a distinction you only learn by missing a date that belonged to somebody else.
2015 → 2020 Muse (Ultimate Guitar)

The world's largest musician platform.

I was Head of Android for the Ultimate Guitar app and worked on growth product for a platform with 350M+ users. I launched MelodiQ, which recognises music by ear and turns it into something to practise, and built the cross-platform score engine that now lives in MuseScore, the open-source notation editor. Somebody else's company, and still the largest audience anything on this page has reached.
2019 → now Own products

Shipping for myself.

I started shipping on my own account: consumer apps, connectivity, music, messaging, and now AI. Of the 10 in the ledger, 3 were sold and are still maintained by their new owners. I co-founded 8 and founded 2 alone. The first two shipped while the Muse job was still running, which is what starting out looks like from the inside.
2021 → now Nicegram

Five years inside Telegram.

Nicegram began as a hypothesis about selling info-products inside Telegram, which is not what it became. Several pivots later it was a client with 56M+ organic installs in five years, reached through SEO, referrals and viral mechanics rather than paid acquisition. Nicegram Business is the next turn of the same track, in early access at nicegram.ai.
Now Warsaw, Poland

AI products, and the standard under them.

I build AI products and write down how. I wrote The Proof of Done Manifesto for building software when AI agents write the code, and the skills any coding agent can install are its reference implementation. Why publish the method? Because a claim about how somebody works stays unfalsifiable until it is installable, and once it is installable anyone who doubts it can run it against their own repository and see where it holds up.

What it proved.

The bars show relative weight across my own work, not a score against the field. Every row names the artefact behind it, and a row that can't name something you could go and check doesn't appear on the chart.

Mobile products Head of Android at Ultimate Guitar · 3 iOS products built and sold · Nicegram on iOS, Android and desktop
Growth & distribution Nicegram · 56M+ organic installs · no paid acquisition
AI cost engineering My own agent spend · $10,000+/month → ~$1,000/month
AI agents & orchestration Prowl · 448 tools behind one MCP endpoint
Entrepreneurship 13 years · 10 shipped · 3 sold · co-founder on 8
Product vision Nicegram Business · a paid layer over an app that never had seats
Delivery & project management task-pipeline · ten gated stages, brief to acceptance
Market & competitor research Prowl · 33 modules covering SEO, ads, funnels, reviews and pricing
Consumer product at scale Muse (Ultimate Guitar) · Head of Android and growth product · 350M+ users
Monetization Subscriptions · pay-per-call wallets · crypto rails
Funnels & conversion eSIM Plus · paid acquisition to activated traveller
Open source Nicegram GPL-2.0 · CheckMyData MIT · 11 skill packs MIT
Infrastructure PrivateClawd · isolated VM per customer · 60-second deploy

05 Hire

Two ways to work with me.

The duration, what changes hands, who each one is for and its own limit are all on the page, so you can rule either of them out before you write. The price is not, and that is deliberate rather than coy: a printed figure is a commitment that has to be honoured or revised in public, and this page carries only claims it will not have to revise more often than they are true. Both engagements start with one email that names what you want and by when.

Build

2 weeks from brief to deployed

Your hypothesis, live in two weeks.

Bring one hypothesis. Two weeks later you own a working product on a real domain, deployed under your accounts. Not a prototype, not a Figma file, not a deck.
- Scope fixed on day one, because a moving brief is how two weeks becomes two months
- The app, the landing page, payments and analytics wired to each other
- Deployment on your infrastructure, under your accounts
- Source, docs and the agent toolkit that built it, which is the same one this page gives away free, handed over at the end
Who it's for

Founders and teams with one hypothesis worth testing and no appetite for a three-month build.

The limit

One hypothesis per engagement. If two weeks isn't enough, I say so before taking money and tell you what the window does cover.

Send the brief
Training

2 hours live, one session

Find where the agent bill goes.

Two hours, screen to screen, using the setup that took my own agent spend from more than $10,000 a month to about $1,000. We measure what you're spending now, decide what to change, and put the changes in order.
- A spend map built from your real usage, not from the invoice total
- Model routing, context discipline and caching that survive daily work
- The skills and hooks that stop an agent spending tokens on the wrong task
- Before-and-after measurements, so the result is something you can check without me
Who it's for

Solo builders and teams running Claude Code or a comparable agent daily, with a bill that has stopped being funny.

The limit

The 10× result is mine, on my stack, and it isn't a guarantee for yours. The session gives you the method and a measurement at both ends.

Ask about a session

Before any of that there is a free audit: tell me what you have shipped and where you think it leaks, and if the project interests me I go through it and send it to you. I choose which ones I take. Anything else outside these two formats can still start with an email: a longer build, a second look at an agent stack, or a question about whether the thing is worth building at all. That last one is free too, and it has talked more than one person out of a project they were about to pay somebody for. If somebody else is a better fit, that is the answer I give.

06 Writing

Two channels.

I write about building, agents, distribution and what they cost to run. Russian and English are separate rooms rather than translations of each other: a post gets written in the language it was thought in, which is why the two don't line up week to week.

Longer pieces live on the blog, with the evidence for each checkable claim at the bottom of the page.

Follow on X

07 Contact

Reach me.

Email works best. Two lines are enough: what you want, and by when. I'll tell you whether it fits. Terms for the two paid engagements are above, and anything that isn't one of them arrives at the same address anyway.

LinkedIn Sergey Sheleg AI product builder Warsaw, Poland View profile

Follow on X

Sergey Sheleg

Skills for agent work

The ssheleg harness

Project Observatory ↗

Asset Foundry · in development

What's running.

The whole portfolio

How I got here.

Olympiads, then a degree that named the job.

Other people's products, other people's deadlines.

The world's largest musician platform.

Shipping for myself.

Five years inside Telegram.

AI products, and the standard under them.

What it proved.

Two ways to work with me.

Your hypothesis, live in two weeks.

Find where the agent bill goes.

Two channels.

Reach me.