Both put AI at the centre of how you build, but they aren't the same kind of tool. Cursor is a local, AI-first code editor. ShipIt is a self-hosted IDE that runs coding agents — Claude Code or Codex — through your real PR, CI, and preview loop. Here's the honest breakdown.
Cursor is the more mature day-to-day coding surface. It has inline autocomplete and in-editor AI edits that ShipIt simply doesn't offer, years of polish, and a large ecosystem. If you want to stay in a fast editor and steer AI as you type, that's Cursor's job — ShipIt deliberately isn't an editor.
ShipIt is about owning the agent loop on your own infrastructure. It runs whole tasks as parallel container-isolated sessions — each with a live app to test against and a pull request to review — on hardware you control, driving the agent account you already pay for. If that, the GitHub loop inline, and bring-your-own-agent matter more to you than inline autocomplete, ShipIt fits.
| Capability | ShipIt | Cursor |
|---|---|---|
| What it is | Self-hosted, chat-driven IDE that runs coding-agent CLIs through the full ship loop | AI-first code editor (a fork of VS Code) with inline AI assistance |
| Primary interface | Chat + live app preview + inline PR card | A code editor you type in (autocomplete, Cmd-K, Composer/Agent) |
| Hosting model | Self-hosted orchestration — control plane, workspaces, and previews on your VPS or local Docker; no ShipIt cloud in between | Proprietary; orchestration & inference on Cursor's cloud (execution can be self-hosted) |
| Model traffic | Goes to whichever Claude or OpenAI account you configure — self-hosting the stack doesn't make inference local | Routed through Cursor's cloud to its model providers |
| License | Open source, Apache 2.0 | Proprietary / closed source |
| Agents / models | Bring your own agent — Claude Code or Codex, on your subscription or API key; connect several accounts per provider with automatic roll-over before a usage limit stops a session | Frontier models (Anthropic, OpenAI, etc.) + Cursor's own models, routed through Cursor; requires a paid Cursor plan |
| Inline autocomplete | No — not a code editor | Yes — fast tab completion and inline edits |
| Parallel isolated sandboxes | Yes — one Docker container per session, on your infra | Yes — Cloud Agents (managed, or self-hosted workers on your infra) |
| Live app to test against | Docker Compose preview per session, with services the agent starts, stops, and tails itself | Cloud Agents build and test in an isolated worker; in the editor you run your own local dev setup |
| Android build & emulator preview | JDK, Android SDK, and Gradle in the session image; a Compose-declared emulator renders live in the preview panel | You run Android tooling in your own environment |
| GitHub PR / CI / deploy loop | Inline — PR card, CI status, review threads, deploy status in chat; sessions can wake on their own PR merge | Cloud Agents open PRs; CI/review largely handled in GitHub |
| Issue tracking | Linear + GitHub Issues inline; start a session from an issue | Not a built-in issue tracker |
| Mobile | Mobile-first installable PWA, with voice in/out | Desktop editor; Cloud Agents are also reachable from web and mobile |
| Maturity | Early public release; younger, narrower | Mature, widely adopted, large ecosystem |
Cursor is an AI-first code editor built as a fork of VS Code by Anysphere. It keeps the editor you know and threads AI through it: fast tab autocomplete, an inline edit prompt (Cmd-K), and a Composer/Agent mode that can make multi-file changes. More recently it added Cloud Agents that run asynchronously — managed in Cursor's cloud, or on self-hosted workers you run yourself on paid plans. It's a proprietary product — you sign in, and its planning and inference route through Cursor's servers even when execution is self-hosted (a Privacy Mode limits what's retained). It is mature, heavily used, and excellent at the moment-to-moment loop of writing code with an AI looking over your shoulder.
ShipIt is a self-hosted, chat-driven IDE for running coding agents through your real engineering loop. Instead of being an editor, it orchestrates agent CLIs — Claude Code or Codex, on the subscription or API key you already have. Each task runs as a session in its own isolated Docker container, with its own branch, a full clone of your repo, and a live instance of your app (via Docker Compose) to build and test against. The GitHub PR lifecycle — description, CI checks, review threads, deploy status, merge state — renders inline in chat, alongside your Linear and GitHub issues. It runs on a VPS you own or local Docker, is open source under Apache 2.0, and is mobile-first with voice.
Both let you hand work to an AI agent rather than typing every line, and both can run multiple agent tasks in parallel — Cursor through Cloud Agents, ShipIt through isolated sessions. Both can use frontier models from Anthropic and OpenAI. If your question is "can an agent go off and implement this while I do something else?", either tool answers yes.
Be clear-eyed about this. ShipIt is not a code editor — there's no inline autocomplete, no Cmd-K, no typing in a buffer with AI suggestions. If most of your day is hands-on coding in a file, Cursor is simply the better tool and ShipIt won't replace it. Cursor is also far more mature, with years of polish, a large ecosystem, and a deep feature set; ShipIt is in an early public release. And Cursor's hosted model means zero infrastructure to run — you install an app and go, whereas ShipIt asks you to stand up Docker on a machine you own.
ShipIt's bet is ownership and the full loop. The orchestration, workspaces, and previews run on your hardware with no ShipIt-operated cloud in between, using container-isolated sessions, a default-deny outbound firewall per agent, and brokered git credentials. It's agent-agnostic, so you're not locked to one vendor's cloud — and you can connect several accounts per provider, with ShipIt rolling to the next before a usage limit interrupts a session. And it closes the loop end to end: an isolated app to test against with services the agent drives itself, Android builds and a live emulator in the preview panel, the PR/CI/deploy cycle inline down to sessions that wake on their own PR merge, issues from Linear and GitHub in the same place, and a mobile-first UI you can install to your home screen so kicking off and reviewing work from a phone is a first-class path. If the value you want is "delegate a task to my own agent, on my own infra, and review the PR," that's the job ShipIt is built for.
Not exactly — they solve different problems. Cursor is an AI-first code editor you type in, with inline autocomplete and in-editor edits that ShipIt has no equivalent of, plus Cloud Agents for asynchronous tasks. ShipIt is a self-hosted surface for running coding agents (Claude Code, Codex) end-to-end through PRs, CI, and previews. Many people use an editor like Cursor for hands-on coding and a tool like ShipIt for delegating whole tasks to agents.
Partly. Cursor's Cloud Agents do support self-hosted execution — a worker you run on your own machine or VPS (the 'My Machines' option), on paid plans. But Cursor itself is a proprietary application from Anysphere, and even with self-hosted execution the agent's planning and inference still run on Cursor's cloud. ShipIt is open source under Apache 2.0 and runs its orchestration, workspaces, and previews on hardware you control, driving whichever Claude or OpenAI account you already hold — with no vendor cloud in between.
No. ShipIt has no inline tab-completion or in-editor Cmd-K editing — that is one of Cursor's biggest strengths and a clear reason to keep using it. ShipIt's interface is chat plus a live preview and the PR loop, not a code editor you type into line by line.
Yes. ShipIt spawns many sessions at once, each in its own Docker container with its own branch, clone, and live app instance, and each reviewable as its own pull request. Both tools can self-host execution; the difference is that ShipIt's orchestration is open source and runs on your own infrastructure, and it drives the agent account you already hold, whereas Cursor's Cloud Agents keep planning and inference on Cursor's cloud and require a paid Cursor plan.
ShipIt installs with one command and runs on your laptop or a VPS. Bring Claude Code or Codex and the agent you already pay for.
