Skip to content
MakerGain
← Back to cases

RepoPrompt canceled its US$15 monthly fee. Coding agents still need the right files

Edited by MakerGainPublished: Article updated:

RepoPrompt handles selection before a coding agent’s request: models have limited context and attention and often reason after reading only part of a codebase. It assembles relevant files and repository structure into reviewable context for the agent. Eric Provencher later removed the US$15 monthly fee, open-sourced Community Edition, and archived Classic. After joining OpenAI, he stopped actively participating and handed maintenance to the community. The app subscription became free; model calls remain billed under each service’s terms.

RepoPrompt is a macOS app that selects files for coding agents. On July 16, 2024, Eric Provencher sent its first build to testers who had found his TestFlight through a Reddit post about automating copy and paste (Next Chapter[1]). That origin explains the early product: its main path was not fully agentic automation, but human selection of model context to produce targeted, reviewable file edits. Eric later described that form as cumbersome for users, though its techniques still offered clear gains over naive prompting.

The problem remains. When models read repositories, context windows and attention are limited; they often start reasoning after seeing only a small portion of the codebase (Next Chapter[1]). RepoPrompt handles selection before the request. The Community Edition (CE) repository describes assembling files, CodeMaps, repository structure, and Git diffs into focused, reviewable context for AI tools and CLI agents. Its Context Builder lets an agent explore a repository and decide which files matter within a token budget (CE repository[2]). The website summarizes it as organizing context, running deep analysis, and orchestrating coding agents in a native macOS app (website[3]).

Image: official RepoPrompt interface[4], obtained on 2026-11-18, preserving the official interface.

Delegation follows. The website says a reasoning model plans first, then divides the plan into tasks assigned along file boundaries to subagents from different providers. They can run sequentially or in parallel, with the orchestrator checking adherence to the plan before continuing (website[3]). People have not disappeared: context is designed to be reviewable before handoff. The tool takes over searching through files without removing the place for review.

That manual-selection workflow was preserved in Classic, but Classic is archived, receiving neither updates nor support (website[3]). Context Builder pushes selection further upstream. Eric says every in-app workflow prompt depends on it as a brain, and relevant context for most large-codebase problems can fit into about 120k tokens. That is his account, not an independent measurement (Next Chapter[1]).

He also acknowledges complexity as one of the biggest adoption barriers. He tried to compress the techniques into buttons and knobs so users would not need a masterclass to control them (Next Chapter[1]). Context Builder attempts to let the agent perform that selection itself.

In an Indie Hackers interview, he reports commercializing a free tool originally built for himself into five-figure MRR in under a year, with an active power-user community (IH interview[5]). That is an account from the interview, not current revenue or profit. With costs and profit margins undisclosed, MRR cannot establish profitability.

The later Next Chapter announcement makes the pressure more explicit. It became increasingly difficult to persuade people to pay US$15 monthly when they could simply prompt their own agents to solve the problem and it just worked. As a bootstrapped founder, he faced monetization pressure from day one, with subscriptions potentially disappearing the following month (Next Chapter[1]).

In the May 27, 2026 Next Chapter announcement, he says licensing restrictions had been fully removed as of update 2.1.32 and all subscriptions at that time canceled. It also previews a fully free, open-source Community Edition. The June 13, 2026 CE release post then announces its GitHub launch as the home of active development (CE release[6]). CE uses Apache-2.0. To serve coding agents, it removes many old features and focuses on speed, context density, and agent performance. Classic retains the manual file-selection, prompt-assembly, and copy-to-AI workflow, but is archived without updates or support (website[3]).

The easy misunderstanding is what became free: the RepoPrompt app subscription. Its FAQ explicitly says coding-runtime plans, API usage, and other services remain billed under their respective terms (FAQ[7]). Free software does not make model calls free. That spending goes through users' own service accounts.

In Next Chapter, he also says OpenAI's developer-experience lead Romain Huet had invited him to join the team a few months earlier. Before agreeing, his first request was that RepoPrompt's loyal users be looked after (Next Chapter[1]). The CE release says his new job means he will no longer actively participate in the project, with the hope that the community takes over its next generation (CE release[6]).

Together, canceling subscriptions, open-sourcing, and stepping away form a handover of maintenance responsibility. Users get free access and source code; the tradeoff is that the main author is no longer actively involved. Building CE from source requires macOS 26 or later and full Xcode 26. Homebrew installs a signed and notarized public application instead of building from source (CE repository[2]). Open source does not remove those barriers, but ordinary users at least do not need a full Xcode installation first.

RepoNeedle also selects repository context for coding agents. It describes itself as an open-source, token-budget-aware retrieval system that outputs model context through a Python library and command line, rather than a native macOS interface (RepoNeedle[8]). RepoPrompt puts selection, review, and multi-agent orchestration into a macOS app; RepoNeedle makes retrieval a library other tools can call. Both address an upstream problem in different forms. Capability claims come from the projects themselves, without an independent same-task comparison.

RepoPrompt's US$15 monthly fee sold judgment about which files an agent needs in a large repository, rather than a model. Eric later removed that charge and handed maintenance to the community. The file-selection work remains, now without a US$15 monthly subscription supporting it.

Continue reading

AI development
Every small revision came with another fee. He moved app design into a chat box

Paying extra for every revision to a mobile app design led Diego Roshardt to build AppAlchemy. Users describe an app in chat, get React Native code and scan a QR code to preview it live on their phone. A change requested in chat updates the screen. Users can export the code, while the public materials leave it open who packages and publishes the app. Roshardt built the first version in about two weeks. He chose to build a history on Reddit, find relevant communities and share useful posts containing links instead of buying ads.

AI development
Write the instructions before asking AI to code: CodeGuide turned 9 hours of documentation into a subscription

CodeGuide sells an instruction manual prepared for a project: clarify the requirements, architecture, and tasks, then hand them to AI coding tools. It began as CJ’s own workflow for his MVP development service—9 to 10 hours clarifying a project with ChatGPT, then about 7 hours saved by chaining prompts with make.com. Rather than build a complete product first, he used Bolt to create a landing page and collect a waitlist. He individually messaged beta users who commented, asking whether they already had a better workflow for AI hallucinations; his conclusion was that nobody had a better approach than the one he had tried.

AI development
Asked what it could automate, he could only say “anything”: Meerkats later broke that answer into rules

Santanu Dasgupta built Meerkats after two failed attempts. Initially an LLM app builder with a code-free workflow targeting agencies early on, its website now positions it as a context layer for AI systems. Its specific job is to connect business data, definitions, state, rules and permitted actions into managed context for agents to read and act on. The operator has narrowed his old answer of “anything” to execution under rules and permissions, with a receipt left behind.

AI development
Editable app screens from a sentence, and Sleek is up for sale

Mattia Pomelli and two friends built four design tools in a year; their latest, Sleek, took three weeks. Given a sentence, requirements specification, or reference screenshot, it generates consistent, editable iOS/Android screens that users can refine in chat or manually. Pomelli chose to focus on mobile interfaces, with output to native Figma layers, React+Tailwind/HTML, and coding-agent calls. The product is now listed for sale, with listed cumulative revenue of US$196,063 and MRR of US$17,672; profit is undisclosed.