Dev is live · free · last verified 23 Oct 2026 · viability healthy
Dev
dev-3.0 is a free Kanban cockpit for running parallel terminal AI coding agents, each in its own git worktree.
The rare tool that picks a fight worth picking: worktree-per-task plus a board makes parallel agents legible instead of chaotic. v1.60.0's balanced account auto-selection is the quiet win — you stop hand-picking which Claude Code or Codex login still has quota. If you run a single agent, the board is pure overhead. Run three or more and it pays for itself in merge conflicts avoided.
Verdict generatedVerified 8d ago · liveness 72/100 · cite: rightaichoice.com/tools/dev
- Solo developers running 3+ terminal AI coding agents across branches and repos
- Indie founders acting as coder, reviewer and QA who want parallel agent workflows
- Developers comparing the same task across multiple agents, models or prompts
- tmux and terminal power users who want a fleet cockpit instead of dozens of tabs
- Teams that need shared boards, roles or reviewer handoffs — the unit here is the individual
- Developers who want a chat-style graphical agent interface rather than a terminal with tmux
- Anyone expecting a full IDE with debugger, language server or in-app code editing
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip dev-3.0 if you run a single coding agent, want a chat-style graphical agent or a full IDE, or need a shared team board with roles and reviewer handoffs — the Kanban and worktree layer only pays off across several parallel agents on one machine.
The download is $0, but every agent you launch bills you directly — Claude Code, Codex CLI and Gemini CLI are separate subscriptions or API usage under your own provider account.
Free and Apache 2.0 with one edition and no per-seat license, which undercuts seat-priced orchestration tools and Cursor's multi-account agent plans outright. The real spend sits outside dev-3.0: your Claude Code, Codex CLI or Gemini CLI subscriptions, billed per provider. Solo developers and indie founders get the whole feature set for $0; teams needing shared boards and reviewer roles should look elsewhere, not higher up a dev-3.0 price ladder, because there isn't one.
In short
Dev — dev-3.0 is a free Kanban cockpit for running parallel terminal AI coding agents, each in its own git worktree. Best for Solo developers running 3+ terminal AI coding agents across branches and repos, Indie founders acting as coder, reviewer and QA who want parallel agent workflows, Developers comparing the same task across multiple agents, models or prompts. Free to use.
What is Dev?
Dev — dev-3.0 is a free Kanban cockpit for running parallel terminal AI coding agents, each in its own git worktree. Best for Solo developers running 3+ terminal AI coding agents across branches and repos, Indie founders acting as coder, reviewer and QA who want parallel agent workflows, Developers comparing the same task across multiple agents, models or prompts. Free to use.
dev-3.0 (dev3) treats multi-agent coding as a control-plane problem. One Kanban board holds every unit of work — feature, bugfix, refactor — and each card gets a real terminal backed by Ghostty's rendering engine, with tmux preconfigured, sitting inside its own isolated git worktree on its own branch. Parallel agents stop colliding because they were never in the same directory. It's built for the solo developer who is coder, reviewer and QA at once, and already drives terminal agents like Claude Code, Codex CLI, Gemini CLI, opencode or Pi. Hover a card and a live terminal preview refreshes every second; when an agent rings the bell, a red badge appears. Column agents auto-spawn when a task lands in a column, and finished work can pass through a configurable AI Review gate — reviewer agent, model and prompt per project — before it reaches you. Setup runs underneath: copy-on-write cloning pulls node_modules and .venv into new worktrees at near-zero disk cost across 12+ ecosystems, a monorepo filter sparse-checks out only the directories a task needs, and auto-allocated ports arrive as $DEV3_PORT0 and $DEV3_PORT1 so parallel dev servers don't clash. Undecided on an approach? Run the same task 3, 5 or 10 times with different agents or prompts, each attempt on its own worktree, then keep the winner. The v1.60.0 changelog adds managed provider accounts with live Claude Code and Codex sign-in in a terminal window, balanced account auto-selection by 5-hour usage and expiring weekly quota, Reply fields on received review threads, and a read-only file explorer beside task terminals. Also new: named tmux window tabs, rebindable pane-focus shortcuts, and browser mode file open-in-new-tab/download. Free and open source under Apache 2.0 — you pay your agent providers, not dev3.
Key features
- Kanban board with drag-and-drop custom workflow columns
- Isolated git worktree and branch per task card
- Split panes running Claude, Codex, Gemini or opencode in one worktree
- Live terminal preview on card hover, refreshing every second
- Ghostty-rendered terminal with tmux preconfigured per task
- Terminal bell alerts that badge task cards when an agent needs you
- Run the same task 3, 5 or 10 times across agents, models or prompts
- Copy-on-write cloning of node_modules and .venv across 12+ ecosystems
- Auto-allocated ports injected as $DEV3_PORT0, $DEV3_PORT1 into scripts
- Column agents that auto-spawn when a task enters a column
- AI Review column gate with per-project agent, model and prompt
- PR review mode that checks out a remote branch with a structured review prompt
- Spaces for grouping projects, with a rail that filters the dashboard
- Managed accounts with live Claude Code and Codex sign-in in a terminal (v1.60.0)
- Balanced account auto-selection by 5-hour usage and expiring weekly quota (v1.60.0)
Behind the verdict
Parallel agents fail in boring ways. Two write to the same file, one rebases over the other's branch, and your afternoon disappears into conflict resolution. dev3's answer is structural rather than clever: every card gets its own worktree and branch before any agent starts, so isolation isn't something you remember to configure. Pick this if you already live in tmux and run terminals as your daily driver. The Ghostty-rendered panes, per-task session persistence and the ⌘/ shortcut panel tell you the author uses this thing all day. The hover preview is the feature I'd miss most — checking on five agents without clicking into five terminals is a real reduction in context-switching tax. If you want a chat window with an agent and a pretty diff view, pass. There's no in-app code editing and no debugger; dev3 orchestrates terminals, it doesn't replace them. Single-agent users should also skip it — the Kanban layer only earns its keep once you're juggling. Teams looking for shared boards, roles or reviewer handoffs won't find them. The vendor is explicit that the unit is the individual, not the org. That's a limitation, not a philosophy flaw, but plan accordingly. Watch the cost surface. dev3 itself is Apache 2.0 and free with no account, but it drives agents under your own provider accounts — so Claude Code or Codex subscriptions and API usage are billed by those providers. The new balanced auto-selection (picking a login by 5-hour usage and expiring weekly quota) is genuinely useful once you have several subscriptions, though API profiles are excluded from it. Closest alternative is hand-rolling worktrees plus tmux windows, which is free and works fine until you have ten of them. dev3's real addition is the board as an index: labels, full-text search, one window per
Models under the hood
as of 2026-10-25
Limitations
- dev-3.0 ships no model of its own — it launches coding-agent CLIs you already installed (Claude Code, Codex, Gemini CLI, OpenCode, Cursor Agent, etc.) under your own provider account, so subscriptions or API usage are billed by that provider on that provider's terms.
- Integration depth varies: Claude Code, Codex, GitHub Copilot CLI, Oh My Pi and Pi get automatic status updates and session resume; Gemini CLI, OpenCode and Cursor Agent report status themselves (OpenCode resumes only the most recent session); any other CLI tool gets no status hooks, no dev3 skill and no session resume.
- Model selection, permission modes, effort and budgets differ per agent.
- Release builds send anonymous usage analytics until switched off in Settings → System → Telemetry or via DEV3_TELEMETRY=off.
as of 2026-10-13
Researching Dev? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is Dev still active?
Dev is live · free · last verified 23 Oct 2026 · viability healthy.
Viability score
How well maintained and how widely used is Dev? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What's new in Dev
Checked 10 days agoAcross the latest 10 updates: 9 feature updates and 1 changelog entry.
dev3 v1.60.0 adds rebindable terminal pane focus shortcuts
Focus shortcuts work on tmux and native backends including browser; a new shortcut types the tmux prefix so it can sit on a ⌘ key.
dev3 v1.60.0 adds reply fields to received review threads
Each open review thread the agent has received gets a field labeled Reply or Add, so users can answer or extend without editing and re-sending.
dev3 browser mode adds file open-in-new-tab and download
File previews open in a sandboxed tab or download to the remote device; 'Open on Windows' appears when dev3 runs in WSL and the browser is local.
dev3 v1.60.0 adds follow-up prompt box to 'Continue this task's conversation'
Optional prompt typed in + Agent becomes the next message after context retelling, letting a new agent act on new direction with earlier context.
Plain Pi becomes a first-class dev3 harness in v1.60.0
earendil-works/pi receives the dev3 protocol, resumes sessions after restart and moves tasks between columns; missing installs are blocked with installer guidance.
Read-only file explorer added beside dev3 task and project terminals
Explorer lists the task worktree or project folder; click to preview a file, right-click to copy its path or insert it into the terminal.
dev3 v1.60.0 runs Claude Code and Codex sign-in live in a terminal window
Adding a Claude Code or Codex account opens the provider's own login running live on the dev3 host, removing the manual command-copy and Verify step.
Oh My Pi gains managed accounts in dev3 v1.60.0
Settings → Accounts → Oh My Pi runs omp login in a fresh agent directory, supporting ChatGPT, Copilot, Gemini and other providers per launch.
dev3 v1.60.0 replaces tmux bar with named window tabs in task terminal
Task terminals show tmux windows as a tab row, with the agent window marked and pinned first; background windows flag output and bell.
dev3 v1.60.0 adds balanced account auto-selection for launches
Launches can pick 'Auto (balanced)' among Claude Code or Codex subscription logins by current 5-hour usage and expiring weekly quota; API profiles excluded.
What do users say about Dev?
What people actually say about Dev — is it worth it?
Our community sentiment research written Jul 3, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about Dev — pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 23 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified Dev 11 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed pricing page, limitations, models
- — changed who should skip it, summary, our verdict, features, description, pricing tiers, integrations
- — changed who should skip it, pricing tiers, who it suits, integrations, description, features, summary, our verdict, limitations
- — changed switching in, description, use cases, who should skip it, setup time, pricing fit, our verdict, models, switching out, features, integrations, limitations, hidden costs
- — changed limitations, pricing page, platforms, models
- — changed who should skip it, who it suits, description, our verdict, features, pricing tiers, summary
Showing the 6 most recent of 11 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does Dev cost in 2026?
Dev: Free to use. Pricing last verified 23 Oct 2026.
Plans compared
Every published Dev tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Free (open source, Apache 2.0) | Free | — | — |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where Dev's pricing actually pencils out — and where peers do it cheaper.
Free and Apache 2.0 with one edition and no per-seat license, which undercuts seat-priced orchestration tools and Cursor's multi-account agent plans outright. The real spend sits outside dev-3.0: your Claude Code, Codex CLI or Gemini CLI subscriptions, billed per provider. Solo developers and indie founders get the whole feature set for $0; teams needing shared boards and reviewer roles should look elsewhere, not higher up a dev-3.0 price ladder, because there isn't one.
Has Dev's pricing changed?
No plan price for Dev has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick Dev?
Pick Dev if you are Solo developers running 3+ terminal AI coding agents across branches and repos, Indie founders acting as coder, reviewer and QA who want parallel agent workflows, Developers comparing the same task across multiple agents, models or prompts, tmux and terminal power users who want a fleet cockpit instead of dozens of tabs.
Look elsewhere if you are Teams that need shared boards, roles or reviewer handoffs — the unit here is the individual, Developers who want a chat-style graphical agent interface rather than a terminal with tmux, Anyone expecting a full IDE with debugger, language server or in-app code editing, Single-agent users — the Kanban layer is overhead when you only run one agent.
Skip it if Skip dev-3.0 if you run a single coding agent, want a chat-style graphical agent or a full IDE, or need a shared team board with roles and reviewer handoffs — the Kanban and worktree layer only pays off across several parallel agents on one machine.
GeneratedReal-world workflow fit
Concrete scenarios for the personas Dev actually fits — and what changes day-one when you adopt it.
You create a card for a payment-webhook bugfix. Dragging it into In Progress spawns a Claude Code session in a fresh worktree; a second card in another column spawns Gemini CLI on a refactor. Hover previews show both, and a red badge appears when one rings the bell for approval.
Outcome: Two agents work the same repo on separate branches with no merge conflict, and you answer only the agent that actually needs you.
You run the same ticket five times with different agents, models and prompts — each attempt gets its own isolated worktree. You browse the results side by side, keep the winner and delete the rest.
Outcome: You pick the approach by reading real diffs instead of guessing, at the cost of several extra agent sessions on your provider account.
You paste the remote branch name, hit Fetch, open the card and toggle PR review mode. The structured review prompt pre-fills so the agent reviews rather than modifies.
Outcome: You get a structured code review on the contributor's branch without checking it out by hand or risking an agent editing code under review.
Use cases
- Run Claude Code on one branch while Gemini CLI refactors another, all from a single board.
- Run the same ticket 5 times with different agents or prompts, each in its own worktree, and keep the best result.
- Keep a Kanban of bug fixes where each card is its own worktree and check status via the hover preview.
- Launch Codex CLI in one worktree for documentation and in another for test generation without branch conflicts.
- Let terminal bell alerts flag when an agent asks for approval so you stop idling on the wrong tab.
- Check out a remote branch, toggle PR review mode and get a structured review instead of an edit.
- Group client, product and infra repos into Spaces and filter the dashboard to one project space.
- Drag a finished task into the AI Review column and let the configured agent review it before you look.
Setup time & first value
How long it actually takes to get something useful out of Dev — broken out by persona, not the marketing-page minute.
One-person studio on macOS or Windows: install the desktop app, point it at a project and launch your first task in roughly 5–10 minutes — the site ships a guided first-run plus an 'Install with an AI agent' brief. Linux or remote/headless: add time for the dev3 server, a tunnel or systemd service, and per-project setup scripts and port config. Reviewing a PR adds nothing new if the repo is
Switching to or from Dev
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From hand-rolled tmux sessions: keep your zsh prefix or drop it, then replace the tab-per-branch habit with one task card per branch and per-task worktrees.
- →From Cursor's multi-account agents: keep Cursor for editing and hand off one click from the card when you want to change code by hand.
- →From a single terminal tab per agent: create one card per unit of work and let column agents auto-spawn instead of manually cd-ing between branches.
- →From a bare git worktree script: the copy-on-write cloning, ecosystem auto-detect and $DEV3_PORT0/$DEV3_PORT1 injection replace most of the script's job.
- →From a Linear or Jira board: cards map one-to-one to tickets, and dev-3.0 picks up existing remote branches by pasting the branch name.
- ↗To VS Code or Cursor: one click hands the worktree off, and the task's branch stays intact for a normal editor workflow.
- ↗To raw tmux and git worktrees: the board is a layer over your filesystem — every task is an ordinary worktree on an ordinary branch you can keep working in.
- ↗To another agent orchestrator: the source is Apache 2.0 licensed, so you can read how worktrees, ports and the review gate are wired before rebuilding the flow.
- ↗To a team platform: if you need shared boards and reviewer roles, plan a move to a multi-user tool rather than stretching a single-player board.
Integrations
Resources & guides
Tutorials & Learning
YouTube returned 6 videos for “Dev”, and we withheld 6: 6 could not be judged, because “Dev” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Dev.
What are the best Dev alternatives?
The closest alternatives to Dev we list are Cursor, crystl and emdash. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Cursor | Freemium | Pricing not published | — |
| crystl | Freemium | Pricing not published | — |
| emdash | Free | Free | — |
Alternatives to Dev
View all
Tools that pair well with Dev
Common stack mates teams adopt alongside Dev, with the specific reason each pairing earns its keep.
Cursor
Cursor is an AI coding agent that plans, builds and ships features from your editor, CLI, Slack, GitHub and iOS.
crystl
crystl is a macOS terminal for running parallel AI coding agents in isolated git worktrees with floating approval cards.
emdash
Emdash is an open-source agentic development environment for running parallel CLI coding agents in isolated Git worktrees.
Dev vs its closest rival
Dev's closest head-to-head is with Locus Robotics — These tools serve entirely different domains: Dev is for solo developers managing multiple AI coding agents via a Kanban/terminal interface (free, alpha), while Locus Robotics is for warehouses with high-volume order fulfillment using autonomous mobile robots (paid RaaS). Choose Dev if you need structure for parallel AI coding tasks as an individual; choose Locus Robotics if you run a warehouse needing 2-3x productivity gains. Verdict generated
Head-to-head comparisons
Dev vs Locus Robotics
These tools serve entirely different domains: Dev is for solo developers managing multiple AI coding agents via a Kanban/terminal interface (free, alpha), while Locus Robotics is for warehouses with high-volume order fulfillment using autonomous mobile robots (paid RaaS). Choose Dev if you need structure for parallel AI coding tasks as an individual; choose Locus Robotics if you run a warehouse needing 2-3x productivity gains.
GeneratedDev vs Presto Voice
Dev and Presto Voice serve completely different markets and purposes. Dev is a free, alpha-stage desktop tool for solo developers orchestrating multiple AI coding agents via a Kanban/terminal interface. Presto Voice is a mature, enterprise voice AI platform for QSR drive-thrus, focusing on order accuracy and upselling. Your choice depends entirely on whether you need to manage coding agents (choose Dev) or automate fast-food ordering (choose Presto Voice). There is no overlap.
GeneratedDev vs Truleo
Dev is a free, alpha-stage tool for solo developers juggling multiple AI coding agents—ideal for parallel prompt experiments and git isolation. Truleo is a paid, mature intelligence platform for law enforcement, automating lead generation from RMS, CAD, and jail calls. Your choice depends entirely on whether you are a solo dev or a police agency.
GeneratedFrequently asked questions
Best-of guides
Topics
Used Dev? Help shape our editorial sentiment research.