An app with only 1 ratings climbs to #6 on the US Developer Tools paid chart: a $4.99 phone client for an open-source coding agent, how many sales can it make?

ompanion lets you remote-control the open-source coding agent omp on your own computer over SSH, straight from an iPhone. Nothing to install on the computer, $4.99 one-time purchase. In early November it reached #6 on the US Developer Tools paid chart and #1 in Croatia and the Czech Republic, but globally it has only 1 ratings. The rank is real, the volume is small. The playbook behind it, paid native clients for rising open-source tools, is worth studying. omp itself already has no one left to follow.

A post asking for a phone client

In 2026 Sep 10, a user named vuuminhphuoc opened a post on omp's GitHub (oh-my-pi, an open-source terminal coding agent) (issue #11609), titled "Official mobile app, or a web version installable to the home screen with push notifications". He described a very specific scenario: you hand a task to the agent on your computer, and while commuting, away from the desk, or lying in bed, you want to check progress on your phone and give it direction. The official way is opening my.omp.sh in a phone browser, and he listed three problems:

1. When the phone switches to background, the browser freezes the page, and the connection drops and repeatedly reconnects;
2. When the agent finishes, or stops to ask you a question, the phone gives no notification at all;
3. The browser input has no Esc, arrow keys or code-block buttons, so it's hard to use.

16 days later, on Sep 26, Swedish indie dev Edvard Wikhall (edde746 on GitHub) created the ompanion repo, and on the next day shipped 1.0. He had built Plezy before, a $5.99 third-party Plex client that has 124 ratings in the US store and reached #3 on the US Top Paid Entertainment chart, with 4,100 GitHub stars. ompanion follows the same playbook: source code and desktop versions are free and open source (GPLv3), and the iOS version is sold for $4.99.

What it is

The first line of the App Store page is clear: omp normally runs in the terminal on your own computer, and ompanion drives it from an iPhone / iPad over SSH, with no daemon to install on that computer. There's no server to provide, no account to register, and no models included. Models use the account you've already configured in omp.

It addresses nearly every problem in that post, one by one:

• Sessions run independently on the computer, so "even if the network drops or the phone locks, the current run won't stop";
• When a tool call needs your approval, you tap it right in the chat; omp's ask tool becomes a form; and a notification is sent when your quota resets (2.0 newly added);
• You can see word-level diffs, sub-agents, session trees and branches, and all sessions on a machine (including ones opened in the terminal) are listed together;
• It supports jump-host chains and Tailscale, and several devices can connect to the same session at once.

It shipped 5 versions in just over a month. In Nov 4, 2.0 rewrote the whole app from Flutter to the system's built-in WebView, and 2.0 added German, which community contributor goetterbote2342 had been submitting since late October.

How big is it really?

Tiny.

• Ratings: 1 cumulative worldwide (India store). From 10-28 to 11-06 there were only a few more (App Store, 11-06).
• Charts: Developer Tools paid chart. US 10-28 #56,11-02, #8,11-04 and 11-06 are all #6; Czech 11-02 went from #91,11-03 straight to #1; Croatia 11-05 #1, Taiwan #2, Germany 11-02 briefly #2, falling to 61 by 11-06.
• GitHub: 5 stars on the ompanion repo; the installers for five versions have been downloaded 287 times in total (GitHub, 11-06).

Czech went from 91 to #1 in a single day, and on this kind of chart one or two sales can change the rank. 11-06 The US top 10 on this chart are long-standing small tools like Lua IDE, Code App, Source 3, Shell Keyboard and BLE Spy, and the whole chart is small. By this threshold, my estimate is it sells somewhere from single digits to a dozen-plus copies a day (estimate; Apple doesn't publish sales).

Why now

What's really carrying these sales is omp itself growing. omp's GitHub stars were 20,832 at the end of July, 33,552 at the end of September, and 6 38,037 in November, and it's still gaining 80 to 140 stars a day (GitHub). As more people use omp, more people want to check it on their phone. That post is the evidence.

11-02 to 11-05 several countries surged at the same time, which lines up with the 2.0 release and the German version going live. Germany also hit #2 in those days. I can't verify whether they caused it.

Where users come from

I found no sign of promotion: no discussion on X, Reddit or Hacker News, and no media coverage. The only visible sources are these:

1. App Store search (inferred). The name is "ompanion: omp client", so omp users searching for omp in the store see it first.
2. Website ompanion.app and GitHub, with downloads for five platforms. Volume is small.
3. The omp community. The author has raised API-related issues in the omp repo, and ompanion's issues also get real omp users filing reports. Oddly, under that mobile-client post, the authors of both competitors self-promoted, but ompanion did not.

Who it's competing with

It's crowded, and nearly all the competitors are free.

ompanion's paying spot is narrow: pure SSH, nothing installed on the computer, no account, no third-party server, a full GUI, and built only for omp. Among omp users who care about privacy and won't run a relay, it's the only option. But that group is the small slice of omp's tens of thousands of stars who use it daily, use an iPhone, won't install Paseo, and will pay $4.99.

It's also tightly coupled to omp's internals. One user reported that when omp 18.16 renamed an export, the extension ompanion installed on the computer failed to load. omp's official thread already has a "to be evaluated" label; if omp ships its own native app, this gap closes.

If it were you

C: worth learning from, not from omp. omp already has seven or eight mobile clients, mostly free, and even ompanion's detailed build has sold only a few units. Squeezing in means splitting a small paying crowd within an ecosystem of tens of thousands. What's worth taking is the play: find an open-source tool that's gaining traction, wait for its users to ask for a mobile client in their own threads, then ship a native paid client with open source code. The author proved it with Plezy on Plex, but the difference is host size: Plex has tens of millions of users, omp has tens of thousands.

Good fit for: a developer who can build a native iOS (or cross-platform) client and uses an open-source CLI or self-hosted tool daily. You're fine with the first product selling only a few hundred units, and you want to build reputation through an open-source repo, then recoup with a paid version.

Not a fit for: anyone who needs this to pay the rent; anyone who can't ship several releases a month to keep up with upstream changes; anyone whose chosen host already has a free, official, or highly rated mobile app.

How to do it: search the GitHub issues of popular open-source tools for "mobile", "iOS", and "app". Look for threads where users have written a long list of pain points, the maintainers replied "to be evaluated," and few people have offered to build it. The host should ideally have hundreds of thousands of users, not tens of thousands of stars. Differentiate the way ompanion does: pure SSH or direct local connection, nothing to install, no sign-up, no data collection. Developers respond to that. Price the iOS one-time purchase around $5; the desktop app and source stay free.

How to get users: put the host tool's name in your app name ("xxx client") to capture store search. On launch day, reply in the thread asking for a mobile app. ompanion skipped this step, and competitors did it. Ask a contributor in the host community to translate the app; one extra language is enough to rank near the top of a thin country chart and get noticed.

Timing: mobile clients for coding agents have been popping up one after another this year. omp went from zero to seven or eight clients in about half a year (inferred). Pick a host and launch before it ships its own mobile app or a general platform like Paseo integrates it. At this pace, that's roughly 3 months.

What it means you misread the signal: ① The tool you picked has already shipped its own mobile app, or a general-purpose app like Paseo or Happy already supports it; ② It's been a month since launch, the store rating is still single digits, and the Top Paid rank keeps bouncing around. That means the host audience is too small, not that you executed badly; ③ Judge ompanion by its own numbers: if by 2.0 and after the German version its rating is still single digits by year end, then "open source + iOS one-time purchase" can't carry a business in an ecosystem of tens of thousands of users. Next time, pick a host ten times bigger.