HangarAI 2.5.1
New in 2.4: statistics across your machines

A team of coding agents, in one native window.

Hangar runs the Claude Code and Codex (beta) already on your machine, side by side, as many sessions as you need, with a shared board, a shared memory and an orchestrator that hands out the work. Either one is enough. Talk to it in a chat, or watch every terminal at once.

Windows macOS 2.2.0 Linux

Runs locally. No account needed, no telemetry.

HangarAI in its chat view, on a Recipes workspace. You asked Iris, the orchestrator, for a small recipe app; she split it into tasks, started workers, and reports that the data model has landed, that Ben, Dan and Eva are at work, and that Cleo's recipe page passed its check and review. Asked about sharing a recipe by link, she put the question of the domain in Needs you. Beside the chat, the board's tasks with the progress each agent reported, and three things that wait for you; on the right, the inspector shows Iris's session, 15% of her context used and her cost. The same chat view in the GitHub Light theme.
The chat view, on a demo workspace. Every picture on this page is the real app.

Chat view Experimental

Talk to the orchestrator. Not to sixteen terminals.

Switch a workspace to a chat with its orchestrator, Claude Code or Codex. The agent keeps running in the panes behind it, and the panes come back as they were.

  • Tasks beside the chat. Each with the progress its agent reports: a percentage and a short note.
  • What waits for you, on top. A question, a decision the orchestrator handed back, a delivery to merge.
  • Previews and kits too. They open beside the chat as they do beside the panes.

Turn it on in Settings › Experiments.

The chat view's side panel: Tasks, 51% overall, 2 of 8 done. Needs you holds three items: Eva waits for your answer, the task Share a recipe by link with an Answer button, and Cleo's Recipe page with a Merge button. In progress: Ben's recipe list at 60%, Dan's search by ingredient at 40%, Eva's shopping list at 15%, each with its note. In review: Cleo's recipe page at 95%, check and review passed, ready to land.

Orchestrator, workers, deliveries

One plans. The others build. Hangar checks.

Pin a Claude Code or Codex pane as orchestrator: it splits the work into board tasks and starts a worker for each, of either agent, on a branch of its own in its own git worktree. With Claude Code, by default Opus plans, Sonnet works and Sonnet reviews.

The same workspace in the panes view: four Claude Code panes. Iris, the orchestrator, checks Cleo's delivery: merged in a scratch worktree, pnpm test passed, review approved. Ben, a worker on the recipe list, sorts it by name and reports 60%. Dan, on search, wrote an accent-blind normalize function and runs six passing tests. Eva, on the shopping list, waits for your permission to run pnpm add fraction.js; her pane is marked in amber.
Green is an agent at work, amber one waiting for you.

A board the agents share

Tasks with the files they own and what they depend on, claimed atomically over MCP. Hangar never runs two tasks at once whose files overlap.

Two columns of the Recipes board. To do: a print-friendly recipe page, blocked by #t3, the recipe page; Share a recipe by link, labelled needs-you and blocked on a question, which domain the links should use. In progress: the recipe list by Ben, search by ingredient by Dan, and the shopping list by Eva, who waits for you.

Every delivery is checked before it lands

A finished worker hands in its branch. Hangar merges it in a scratch worktree, runs your check command and has a reviewer agent read the diff. A step that fails goes back to the worker.

Delivery of #t3 Recipe page
  1. Merged onto main in a scratch worktree Your checkout is left alone until the fast-forward.
  2. pnpm test passed The check command you set, run by your shell.
  3. Review approved A reviewer with a fresh context, read-only tools.
  4. Landed: main fast-forwarded Local only. Nothing is pushed.

An illustration of the stages, with the task in these pictures.

Hangar never pushes

Merges are local fast-forwards. Pushing, pull requests and releases stay your call.

Checked in batches

Deliveries that wait together share one run of the check. If it fails, Hangar finds the one that breaks it and sends back only that one.

On the worker's pane

Where its delivery stands: queued, merging, checking, reviewing, landing, then how it ended.

Woken for what matters

The orchestrator rests while workers run. A delivery, a question or a worker that stops wakes it at once; the rest comes in a digest, every five minutes at most.

No false alarms

A worker busy with a long build or test is not reported idle, and one that is idle is reported once, not over and over.

Rules a prompt can't bend

Permissions, the check command and the branch work lands on are set per workspace and enforced by Hangar.

Hangar Mobile

Answer your agents from your phone.

See what waits for you in every project, answer your agents' questions with a tap, allow or deny a permission. Hangar serves the app itself, over your own Tailscale network: no app store, no account, no server of ours.

Hangar Mobile's For you screen: four things wait. Waiting for your answer, Ben and Cora in hangar-ai, for 4 and 2 minutes, each with an Answer button. Questions for you: should Tailscale Funnel stay out of scope, or be offered as an option? Below, the tab bar: For you, Projects, Shared and Manager.
For you. What waits for you, in every project.
Cora's conversation, a Claude Code pane, waiting for your answer: her summary of the checks she ran, then a card, Cora asks you 3 questions. The first, shall I submit the delivery now, offers Submit now, Wait for the screenshots and Other as large buttons, with Next.
An agent asks. Its questions as buttons.
Ben's conversation: the files he read and edited, one line each, then a card, Ben asks for your permission. Bash, run the window ops' tests, with the command pnpm vitest run src/lib/remoteControl.test.ts, and the buttons Deny and Allow.
A permission. What it will run, Allow or Deny.
The Shared screen: a picture Cora shared from hangar-ai an hour ago, the For you screen in dark mode, with an Open button; under it, lighthouse.html, a Lighthouse report Mia shared from storefront.
Shared. The images and files agents send you.

Hangar Mobile running on sample projects.

Private, through Tailscale

Hangar Mobile listens on 127.0.0.1 only: Tailscale, a free app that links your own devices, is the only way in, over HTTPS. Each phone pairs with a one-time code. Notifications reach it even with the app closed, sent by your PC.

Set up in about a minute

  1. Install Tailscale on the PC and the phone, on the same account.
  2. In Hangar, Settings › Mobile: turn on Hangar Mobile, then HTTPS access.
  3. Scan the QR code, add Hangar to the Home Screen.

Kits & previews

Agents show you their work, beside the terminal.

Over MCP, an agent opens images, code, Markdown, PDFs, web pages, videos and 3D models next to its pane, then renders snapshots to check its own work. Kits bring skills, MCP servers and tabs of their own.

Eve, a Claude Code pane in a landing-pad workspace, was asked to show public/models/hangar.glb and say what it is. She called model_show and model_snapshot over MCP; beside her terminal, Hangar's 3D preview shows hangar.glb, opened by Eve: an arched hangar with yellow doors on a landing pad marked H, with cones, a windsock, crates and a quadcopter. Under it, its Flight loop animation and the model's figures: 4,432 triangles, 32 materials.
3D that checks itself

glTF, GLB, OBJ, STL, PLY and FBX. A model reloads when its file changes, and glTF animations play.

Kits with native tabs

A kit's tab is an HTML page, or a native one in Rust compiled to WebAssembly that Hangar draws in your theme.

33 themes, dark and light

They restyle the whole window, not only the terminals. Graphite is the default, where colour only marks a state.

Try one on this page

Experiments Off by default

Try what comes next.

Under Settings › Experiments. Unfinished work: they can change, break or disappear in any update.

Nodes: Hangar on another machine

Add a Linux machine you reach over SSH: Hangar installs a headless hangar-node, starts it and keeps a tunnel to it. A workspace can live there, its terminals, agents and board included, and the window shows it like a local one. It keeps working there with your computer off.

Or send one worker to the node for a heavy build or a GPU box: its branch comes back to be checked, reviewed and landed here.

Settings › Nodes: Linux machines Hangar reaches over SSH with your keys. The node gpu-box is Connected: hangar-node 2.2.0-rc.1 on gpu-box (x86_64-unknown-linux-gnu), with git and without Claude Code, with Disconnect and Remove buttons.

Chat view

A workspace as a conversation with its orchestrator, its tasks beside it. See above.

Pane scopes Linux

Each pane in a systemd scope of its own: when memory runs out, the system stops the pane that uses it, a build say, not Hangar and every agent.

Performance

Native, in Rust. And lighter with every release.

No Electron: Hangar draws its own window, and each terminal reads its output on a thread of its own. Measured, not estimated.

2.2, against the release before

2.9%

Hangar's CPU with eight agents printing, chat view

was 8.1 %

3.2s

Eight sessions all back after a relaunch

was 9.1 s

0.38s

A workspace in the chat view, ready

was 0.66 s

79MB

The executable, with whole-program optimization

was 119 MB

2.0 native, against 1.12.1 in a web view

249MiB

16 terminals printing non-stop for 30 s

1.12.1: 8.1 GiB

2.5ms

Echo of a key in a pane while 15 others print

1.12.1: no echo within a minute

0.57s

Cold start, four prompts readable

1.12.1: 0.96 s

Where they come from: 2.2's, in the changelog · how 2.0 against 1.12.1 was measured

Also in the box

A shared memory

Notes every agent can search and write, across workspaces: conventions, decisions, traps.

Claude Code and Codex Beta

Either one is enough. An orchestrator of either agent hands tasks to workers of either agent, and each workspace keeps an account per agent.

Delegate, and images from Codex

An agent hands a sub-task to the other and gets the result back. What Codex draws shows in the chat, on your phone and on Discord.

Up to 16 panes per workspace

Claude Code, Codex or any shell: PowerShell, cmd, WSL, zsh, bash, fish, Nushell.

Sessions come back

Reopen the window and every agent pane resumes its conversation, the ones on screen first.

Idle workspaces sleep

After 10 minutes untouched and out of sight, a workspace stops its agents to free memory, and resumes them when you return.

Where the tokens go

Tokens, active time and an API-equivalent cost by project, model and agent, read from your agents' own transcripts.

Dictation

Hold a key and speak: whisper.cpp transcribes on your machine. Off until you turn it on.

Broadcast

One instruction to every pane; agent panes can hold it until their agent is free.

A project manager

One agent that follows every workspace: ask where they stand, or pass an instruction to an orchestrator.

Your agents, set up for you

When neither Claude Code nor Codex is there, one click installs either with its maker's official installer; then you sign in.

Any file, in the viewers

Images, PDFs, code, videos and 3D models from anywhere on the computer, not only the project. A viewer loads only what was shown.

Hangar on Discord

Experimental: a server of your own as a remote, a channel per workspace and a thread per worker, with the files agents share. ⚡ sends a waiting message at once.

An account, if you want one

Sign in with GitHub to see the statistics of all your machines together. Off by default; what it stores.

Download HangarAI 2.5.1

Install it once: it offers each new version and installs it in one click.

Windows

64-bit installer, for you alone, no administrator rights.

Download .exe

Not code-signed yet: SmartScreen asks once. What to do

macOS

2.2.0, the latest for the Mac for now. Universal, macOS 10.15 or later.

Download 2.2.0 .dmg

New macOS builds are on their way. Details

You also need Claude Code or Codex (beta), or both: Hangar installs them if they are missing, and you sign in once. Every file of the latest release

Questions

What does it cost?

Nothing at the moment, and no account is needed. The agents cost what they already cost you, on your own plan or API key. An orchestrator, its workers and a reviewer run at once, so they use a plan's limits faster than a single agent.

Does it work with my Claude or ChatGPT plan?

Yes. Hangar starts the claude and the codex you installed, with the logins they already have, and never asks for your credentials. The navigator shows your plan's five-hour and weekly usage.

Why is Codex marked beta?

Hangar drives Codex through the protocol OpenAI publishes for it, and was written from that protocol's schema, source and docs rather than from recorded sessions. Until Codex users confirm it, it stays beta. Something wrong in a Codex pane? Report it. What works with which agent: the Codex guide.

Does my code or data leave my machine?

Not through Hangar: no telemetry, and your code, prompts and transcripts never reach a server of ours. The board lives in your project (.hangar/board.json); memory, settings and statistics stay in the app's data folder. Hangar goes online to check GitHub for updates, for the dictation model or Groq if you turn them on, for Hangar Mobile's notifications and Discord once you enable them, and for the optional Hangar account once you sign in.

Do I need a Hangar account?

No. It is off by default, and Hangar works the same without one: it never contacts the account server until you sign in. Sign in with GitHub, in Settings › Hangar account, to link your machines; turn on the statistics sync, a switch of its own, and the Statistics page shows all of them. It never receives a file path, a prompt or a line of code. What it stores, and how to delete it.

Will the orchestrator push to my repository?

No. Workers commit on branches of their own, and a delivery lands on yours as a local fast-forward. Hangar never pushes, opens a pull request, tags or deploys, and its agents are told not to either.

Why does Windows warn me about the installer?

It is not code-signed yet, so SmartScreen warns the first time: choose More info, then Run anyway.

Is there a version for the Mac?

Releases after 2.2.0 are built for Windows and Linux only. 2.2.0's universal disk image, signed and notarised, is the latest one for the Mac for now, and there is none to update it to yet. New macOS builds are on their way.

What does Hangar Mobile need?

Hangar open on a PC that stays on, and Tailscale on the PC and the phone, on the same account. On an iPhone, notifications need iOS 16.4 or later and the app opened from the Home Screen. Merging a delivery is still done on the PC.

Where do I ask a question?

On the Discord. Bugs and requests go to GitHub issues.