EN
Back to the archive

The archive · Developer & Business Tools · Product decision · 2022–2026

Orca's parallel-agent ADE bet: +27k stars in July 2026, 53k+ by August

Stably AI's free MIT 'agent development environment' runs Claude Code, Codex and friends side by side in isolated git worktrees; #2 star growth in July 2026.

Stably AI (Orca)

The betThat the CLI-agent era's bottleneck is orchestrating many agents at once, so a free MIT bring-your-own-agent ADE with git worktrees beats single-vendor IDEs.Live

What the business is

Orca is an open-source 'agent development environment' (ADE): a desktop app that runs many CLI coding agents — Claude Code, Codex, Cursor CLI, OpenCode and others — in parallel, each in its own git worktree, with diff review, task boards, design-mode feedback, SSH remote runs and a mobile companion; users bring their own agent subscriptions.

How it started

Stably AI was founded in 2022 in San Francisco by Jinjing Liang (CEO) and Neil Parker, and went through Y Combinator's W22 batch; the YC directory describes the four-person company as makers of Orca, an MIT-licensed terminal-based agent orchestrator. The Orca repository was created in March 2026, riding the moment when developers began running multiple CLI coding agents and paid for several subscriptions at once.

What happened

Orca grew through GitHub in 2026: the yeyupiaoling write-up (2026-07-28) dates its surge to July weekly trending (following Joao Queiros' 2026-07-15 roundup) and puts stars above 30k in late July, praising the isolation of parallel git worktrees, SSH worktrees for remote machines, Design Mode for annotating UI feedback, and iOS/Android companions. OpenGithubs' 2026-07-31 monthly rank lists 33.7k stars and +27,050 growth for July, second only to DietrichGebert/ponytail in the same list. Habr (2026-08-24) reports more than 53k stars by August 25 — under six months from creation — and details the free MIT desktop app with 35 preset agent configurations, per-agent quota views, workspace boards, and no ADE subscription: users pay only their agents' own subscriptions.

How it ended up

Still operating as of 2026-09-02: the repo sits above 53k stars per Habr's August reading, the desktop app ships for macOS, Windows and Linux with mobile companions, and Stably AI remains an active YC company. Monetization beyond a free MIT tool is not public.

Background

Orca is an open-source 'agent development environment' built by Stably AI, a Y Combinator W22 company in San Francisco. Where an IDE is designed for a human writing code, Orca is designed for a human directing a fleet of AI coding agents: Claude Code, Codex, Cursor CLI, OpenCode and others run side by side, each in its own git worktree so parallel agents never clobber each other's changes, with diff review, task boards, per-agent quota visibility, remote SSH worktrees and a mobile companion.

Stably AI was founded in 2022 by Jinjing Liang (CEO) and Neil Parker, and the Orca repository appeared in March 2026, when developers were subscribing to multiple CLI agents at once. Orca's bet is bring-your-own-agent: the MIT app itself is free and charges no subscription, and users connect their own agent accounts; official presets cover more than 25 to 35 agents, and any agent that runs in a terminal can be added.

The growth curve is steep and dated: OpenGithubs' July 2026 monthly rank records 33.7k stars with +27,050 added in July, second-fastest among tracked open-source projects, after weekly-trending appearances that month; Habr reports over 53k stars by 25 August 2026, under six months after the repo was created. How a free, open, model-agnostic tool converts that mindshare into revenue remains unanswered as of 2026-09-02.

What has to be true

  • The 2026 explosion of CLI coding agents and multi-subscription workflows created demand for a neutral orchestrator rather than one more agent or IDE.
  • Bring-your-own-agent plus per-agent git worktrees solved the concrete pain of parallel agents overwriting each other and losing context.
  • Free MIT licensing and no model lock-in lowered the switching cost to zero against Cursor- or Copilot-style integrated tools.
  • Parallelism-first productization — fan-out, diff annotation, Design Mode, mobile monitoring — turned a developer workflow into a visible product category.

What can be applied

When an ecosystem fragments across many fast agents, the neutral orchestration layer that runs them all wins stars; free and open gets distribution, but monetization waits until it owns the workflow.

Aftermath

As of 2026-09-02, Orca is live: Habr's 2026-08-25 reading puts the repo above 53k stars, the MIT desktop app supports macOS, Windows and Linux with iOS/Android companion apps, and Stably AI remains listed as an active four-person YC company with no public funding round or revenue disclosed. The open questions are classic for open-core tools: whether parallel-agent orchestration becomes a durable workflow or a feature absorbed by Cursor/VS Code-style IDEs, and how the company monetizes once developers already pay their agents' subscriptions directly.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases