compare
the short answer
If you want to run several coding agents at once and you want it to cost nothing, use Conductor or Nimbalyst. Both are free for local parallel agents, both are shipping, and Nimbalyst is MIT and runs on Windows and Linux as well. Skribbl is $99 a year, macOS on Apple Silicon only, and closed source. That is the honest opening and everything below it is detail.
What Skribbl has that none of the tools on this page appear to have is two things. A spend meter that is always on screen, so you are never one invoice away from finding out what a fleet costs. And drawing as the governance primitive: a line you draw between two agents is what permits one to command the other, and erasing the line revokes it. Everything else here isolates agents from each other. Skribbl is about what happens when you deliberately connect them.
the table
Checked 2 August 2026 against each tool's own site, repository and release feed. Two of the best known entries are no longer maintained, which is worth knowing before you spend an evening on one.
| tool | shape | platforms | price | source | agents | spend meter | worktrees |
|---|---|---|---|---|---|---|---|
| SkribblPre-1.0, shipping | Infinite canvas | macOS, Apple Silicon only | $99 a year for the first 25, then $99, or $29 a month. One day free with no card, then a card unlocks 13 more | Closed | Claude Code, Codex, Gemini, Grok | Yes, in the top bar at all times. Claude Code only today | Yes, bound to a group |
| ConductorShipping. Built by Melty Labs | Workspace list | macOS | Free for local parallel agents. $50 a month for cloud workspaces | Closed | Claude Code, Codex, Cursor, OpenCode | Not documented | Yes, one branch per workspace |
| Vibe KanbanBloop shut down 10 April 2026. Last release 24 April 2026 | Kanban board | Anywhere Node runs, via npx. Browser UI | Free | Open, Apache-2.0 | Ten or more, including Claude Code, Codex, Gemini, Copilot, Amp, Cursor | Not documented | Yes, with cleanup of orphaned ones |
| CrystalDiscontinued, replaced by Nimbalyst. Last release 26 February 2026 | Session list | macOS universal, including Intel. Linux | Free | Open, MIT | Claude Code, Codex | Could not verify | Yes, central to it |
| NimbalystShipping actively | Visual workspace and session board | macOS including Intel, Windows, Linux | Free | Open, MIT | Claude Code, Codex | Not documented | Yes, optional per session |
| Claude SquadShipping actively | Terminal interface | Anywhere Go runs | Free | Open, AGPL-3.0 | Claude Code, Codex, OpenCode, Amp, Aider | Not documented | Yes |
| SculptorIn beta, built by Imbue | Session list | macOS Apple Silicon, Linux | Free while in beta | Could not verify | Claude Code | Not documented | No. Containers, with a full repository copy each |
scroll the table sideways for spend meter and worktrees
vs Conductor
A macOS app that runs several coding agents in parallel, each in its own git-worktree-backed workspace with its own branch, terminal and diff.
where Conductor wins
- It is free for the thing most people come here to do. Unlimited local parallel agents cost nothing, with no card and no trial clock, against ours, which starts at $99 a year. That is the sharpest difference on this page and we are not going to bury it.
- It is a funded company shipping at 0.78.x, and "will this still exist next year" is a fair question to ask a one-person pre-1.0 app.
- Cloud workspaces, multiplayer and an API. Agents that keep running when the laptop shuts. Skribbl is local-only and has no answer to any of that.
- It supports Cursor and OpenCode, which we do not.
where they differ from us
vs Vibe Kanban
A kanban board for planning work and dispatching it to coding agents, each task running in its own isolated workspace with a branch, a terminal and a dev server.
where Vibe Kanban wins
- Apache-2.0 and self-hostable. It runs on any operating system, there is no vendor to lose, and it can be forked forever. Skribbl is a closed macOS binary with a licence key.
- A backlog you groom and dispatch from is a genuinely different and better model for queued work than a canvas of live terminals. If your work arrives as tickets, this shape fits it and ours does not.
- It drives more agent CLIs than anything else here, and pairs inline diff review with a built-in browser preview that has device emulation.
where they differ from us
vs Crystal
A desktop app for running several Claude Code and Codex sessions at once in parallel git worktrees, to try an approach more than one way and compare the results.
where Crystal wins
- MIT, free, and readable. You can check what it does to your repository rather than take a stranger’s word for it.
- It runs on Intel Macs and on Linux. We ship one build, for Apple Silicon, and that rules out a real number of people.
- Running the same task several ways and diffing the outcomes is a workflow it serves directly. Skribbl is built for a fleet doing different work, not the same work in triplicate.
where they differ from us
vs Nimbalyst
Crystal’s successor: a visual workspace holding markdown, diagrams, mockups and a browser preview beside a kanban of agent sessions.
where Nimbalyst wins
- It is free, MIT, and on all three desktop platforms. Of everything on this page it is the one closest to what we are building, and it costs nothing.
- A visual workspace that holds diagrams, mockups and documents alongside sessions. Skribbl’s canvas holds terminals and files, not artefacts you author.
- Mobile companions. We have none and none planned.
where they differ from us
vs Claude Squad
A terminal program that manages several agents at once in isolated worktrees, without leaving the terminal.
where Claude Squad wins
- It never asks you to leave the terminal, which for a lot of people is the entire requirement and the reason a desktop app is a downgrade.
- It is free, AGPL-3.0, and runs over SSH on a machine with no display at all.
where they differ from us
vs Sculptor
A desktop app that runs parallel Claude Code agents in Docker containers, each with a full copy of the repository, and syncs any agent’s work into your editor live.
where Sculptor wins
- Containers are stronger isolation than worktrees. An agent that runs a destructive command hits a container rather than your machine, which is a real answer to a real fear that we do not have.
- Its pairing mode pulls whichever agent you are watching into your own editor live.
- Free while it is in beta.
where they differ from us
the editors: Cursor, Copilot, continue.dev and VS Code
Everything above is a tool for running several agents at once. The four below are not: Cursor, GitHub Copilot and continue.dev are single-agent, in-editor assistance, VS Code is the editor those extensions run inside, and between them they are what most people actually have open all day. They are on this page because they are what people search when they are looking for something better, and it would be dishonest to leave them off and let you assume we compete with them.
We do not. Skribbl is not an editor and does not compete on autocomplete. It has no completions, no inline edit, no language server, no extensions, and it ships no model of its own. It is the layer above the editor: several agents running as real terminals on one canvas, authority between them granted by a line you draw and revoked by erasing it, each one isolated in its own git worktree if you want, and a meter in the top bar that says what the whole fleet is costing right now.
Skribbl vs Cursor
An AI-first code editor, built as a fork of VS Code, with predictive tab completion, inline editing by instruction, and an agent that can plan and edit across a whole codebase.
where Cursor wins
- Tab completion and inline editing are the thing it is best in the world at. It predicts the next edit, not just the next token, and it applies a change where your cursor already is. Skribbl has no editor and no completion of any kind, so if that is the loop you live in, this is not a trade you should make.
- It is a full editor, so it does everything an editor does: language servers, extensions, debugging, go-to-definition, all inherited from VS Code. Skribbl assumes you already have an editor open and does not try to be one.
- It runs on macOS, Windows and Linux. We ship one build, for Apple Silicon.
- It has a free tier, so you can find out whether you like it without a card. Ours is one day free, then a card.
- Its background agents run work off your machine, which we cannot do at all. Anyone telling you Cursor is strictly one-agent-at-a-time is a year out of date.
where they differ from us
Most people looking for a Cursor alternative want a better editor. If that is you, this page is the wrong page: try Windsurf, Zed, or Copilot inside the editor you already know. The people Skribbl is for keep Cursor open and run it beside us, because the question we answer is what four agents are doing at once and what they are costing, and no editor answers that from inside a single window.
Skribbl vs GitHub Copilot
Inline code completion and chat inside the editor you already use, plus an agent mode and a coding agent on GitHub that can take an issue and open a pull request.
where GitHub Copilot wins
- Ubiquity. It is in VS Code, Visual Studio, JetBrains, Neovim and Xcode, which means it is in the editor you already have rather than asking you to adopt a new one. Nothing else on this page is that easy to start.
- It is the cheapest way into serious AI assistance. There is a free tier with monthly caps, and the paid individual plan costs a fraction of a standalone agent subscription. Skribbl is a yearly licence on top of whatever you already pay for a model.
- Enterprise reality: org-wide policy controls, licence management, audit, and a procurement path your employer has probably already walked. We are a one-person pre-1.0 app with a licence key and no team features at all.
- Its coding agent lives on GitHub, so work happens against an issue and lands as a pull request that your existing review process already handles. Skribbl produces branches on your own machine and nothing else.
- It is deeply wired into the rest of GitHub: pull request summaries, code review suggestions, and the repository context that comes with being the platform your code is already on.
where they differ from us
A GitHub Copilot alternative search is usually about completions, price or which models you get, and on all three of those Skribbl is not a competitor at any price: we have no completions and we ship no model. What we are is the layer above. Copilot helps you write the line. Skribbl is where you watch six agents work in parallel, decide which of them is allowed to command which, and stop the one that is burning money.
Skribbl vs continue.dev
An open-source AI code assistant for VS Code and JetBrains that you point at whichever model you like, including models running locally on your own machine.
where continue.dev wins
- It is open source. You can read exactly what it sends where, fork it, and keep it alive if the company behind it stops. Skribbl is a closed macOS binary and you are taking our word for it.
- Model flexibility is its whole design. Bring your own API key, mix providers per task, or run entirely local models so no code leaves the machine. Skribbl drives four vendor CLIs and nothing else, so if your employer forbids sending code to a third party we have no answer and continue.dev does.
- The extension itself costs nothing. You pay whatever your model costs and no more. Skribbl adds a yearly licence on top of that.
- It is configurable to a degree we are not: your own rules, your own context providers, your own slash commands, all in files you check into the repository.
- VS Code and JetBrains, on every platform those run on.
where they differ from us
If you came looking for a continue.dev alternative because you want local models or you want to read the source, Skribbl is the wrong answer and we would rather say so here than waste your evening. Come back when the problem is that you have three agents running and no idea which one is stuck.
Skribbl vs VS Code
A free, cross-platform code editor from Microsoft, built on an open source core, where AI assistance arrives through extensions rather than being the product itself.
where VS Code wins
- It is free, and its core is open source under the MIT licence as the Code OSS repository. Skribbl is a closed binary with a yearly licence key, so on both price and openness this is not a close call.
- The extension marketplace is the reason it is the default editor: language support, linters, formatters, debuggers and themes for practically anything you will ever open. Skribbl has no extension model at all and is not going to grow one soon.
- It runs on macOS, Windows and Linux, on Intel and on ARM, and it also runs in a browser. We ship one build, for Apple Silicon Macs.
- It is the host for the AI tools people actually use, including the official Claude Code extension and GitHub Copilot. If you want an agent inside the file you are editing, this is where it lives, and we do not compete for that slot.
- Remote development is a genuine advantage we have no answer to: it edits over ssh, inside a dev container, or in WSL, with the language servers and the terminal running on the remote machine. Skribbl runs agents on your own Mac and nowhere else.
where they differ from us
There is no real contest here, and pretending otherwise would be silly. VS Code is where you read and edit code. Skribbl is where four agents run as live terminals with the authority between them drawn as lines and the spend on screen. Most people using Skribbl have VS Code open on the other half of the display, and that is the arrangement we designed for rather than one we tolerate.
where we win
Three things, and they are narrow on purpose. A page that claimed we win everywhere would be worth nothing to the person reading it.
drawing is the authority
Every other tool here isolates agents: a worktree each, or a container each, and no way for one to reach another. Skribbl lets one agent command another, and the permission to do so is a line you drew. It is checked in the router, not in the interface, so an agent cannot talk its way past it.
$ skribbl dispatch --to all-children "rerun the suite" Sent to tests, docs.
Erase the line and the next dispatch is refused. There is no permissions screen, because the board is the permissions screen.
the meter is always on
A ten-block gauge in the top bar, at all times, that you never have to click. On API auth it reads dollars and can hard-stop a run you started; on a subscription it reads headroom and only ever tells you. No tool on this page documents anything equivalent.
everything on one surface
Terminals, agents, files, diffs, git history and a live web preview are all nodes on one plane, at whatever zoom you want. A session list shows you one thing at a time and a kanban board shows you cards about things. This is a real difference in how the work feels, and it is also entirely a matter of taste, so try it before you pay for it.
how this was checked
Every row was read off the tool's own website, its repository, or its latest release on 2 August 2026. Nothing here is quoted from a roundup post. Where a fact could not be confirmed from a primary source, the cell says "could not verify" rather than guessing in whichever direction flattered us.
The four editors are held to a stricter version of the same rule. Their plans change more often than anything else on this page, so their price rows describe the shape of the pricing and send you to the vendor for the figure, rather than freezing a number here that would be wrong within a quarter.
Prices and platforms move. If you are reading this months later, check the tool's own page before deciding, and treat the date above as the expiry on everything here.
Our own row is on the same terms: pricing has the full ladder, the docs list what is unfinished, and the questions page answers the ones people actually ask, including the awkward ones.