Stacks/Coding agent
Shopify River on Remote Dev (mobile)
Shopify gave its Slack coding agent River a pool of warm remote Macs with Android emulators, so many agents can build, test, and record iOS and Android changes in parallel. A skill makes River prove its work with screenshots, Live Preview lets anyone try the running app, and a cheap classifier grades conversations to improve River. River went from 6% to 61% of merged mobile PRs.
- Use case
- Shopify's Slack agent River building, testing, and proving mobile (iOS and Android) changes on remote Macs, from bug fixes and PR feedback to overnight automated fixes
- Category
- Coding agent
- Team size
- not stated
- Orchestrator / framework
- River runs on Linux and claims a warm remote Mac (main checked out, apps built, caches warm) with an attached Android emulator on a nested-virtualization Linux VM, over SSH. A CLI wraps the SSH flags; long tasks run in tmux panes River can watch. A skill teaches River to test changes and prove its work with screenshots and recordings. Macs are destroyed after each session and idle ones rotate.
- Memory
- not stated
- Evals
- Jev classifies every mobile River conversation and grades outcomes; failures feed an LLM that proposes improvements, run iteratively through River itself. Adoption tracked as River's share of merged mobile PRs (6% to 61%).
- Observability
- All River conversations are visible (multiplayer by default) and classified by Jev to find where people get stuck or tool access is rejected.
- Cost (stated in writeup)
- Not stated
- Cost per output (stated)
- Not stated in the writeup
- WhatWeRun estimate
WhatWeRun estimate
Not enough info to estimate
The post doesn't name River's model or give tokens or sessions per month.
Reviewed Nov 13, 2026
- Observability tools
- not stated
- Other tools
- SlackRemote Mac fleetiOS Simulator / Android EmulatorLive Preview (WebSocket device streaming)tmux
- Source
- https://shopify.engineering/remote-dev
Michael Garfinkle, Mauricio de Meirelles, and David Cortes · Published Oct 22, 2026 · Last verified Nov 13, 2026
Related
More like this
corezoid.comCorezoid's ticket-routing coding agent
A coding agent triggered by Jira labels that fixes tickets and opens merge requests, with a digital twin of the engineering org deciding which repository, process, or service each ticket belongs to
Stated: ~$2.50 per code change+1
aboutamazon.comThe agent teams that build AWS Emulator
Autonomous agents that write, test, and fix every line of AWS Emulator, a downloadable local emulator of AWS services, with no human-written or human-reviewed code
blog.duolingo.comDuolingo AI STR Fix bug agent
Investigating and fixing Shake-to-Report bug tickets: gathering context (relevant code, tracking events, experiment states, recent PRs, repro steps) and proposing a fix for engineers to verify