One window for your Claude Code and Codex agents

Run a dozen agents side by side and see which one waits for you. Fleet also shows where every token and dollar went.

Free for 3 panes. Pro is $24 at launch, paid once. Apple Silicon, macOS 13 or later.

Before
After

The cockpit with every agent and its cost

Fleetacme · 6 panes
Restart to apply
5h 41%week 23%
Sample data 2 waiting

Fleet shows which agent needs you

Each pane shows WAITING, DONE or working, with its cost and context, read from Claude Code's own hooks. A sound, a Dock count and ⌘J take you to the next one.

Worktree screens
Each screen gets its own branch and worktree, so two agents never edit one file.
Smart compact
Compacts after a chat finishes and waits for you, never mid tool loop. Opt-in.
Chats come back
Each pane resumes its own conversation by id, also after you quit Fleet.
Skills panel
Search every skill and plugin command and send one to the pane.
Many panes, flat memory
In our test, 36 panes streaming heavy output kept Fleet under 400 MB, and its use fell after the first burst. Each agent's own process is extra.

See what each agent changed, and what it left on disk

Fleet shows the files each agent touched, opens its folder in your editor, and finds old worktrees and caches that can go to the Trash.

Files panel
The pane's folder as a tree. Files the agent changed carry a dot, with a diff and a Quick Look preview. Read only.
Open in editor
One click opens the pane's folder in VS Code, Cursor, Zed, Xcode or Finder.
Disk map
Sizes of agent worktrees, build folders and Claude Code caches. A dry run comes first, and items go to the Trash. Transcripts and worktrees with unsaved work are never offered.
One-click updates
Fleet installs a new version and restarts. Each pane resumes its own chat.

Where your tokens and dollars go

Fleet reads your own transcripts on your Mac. Nothing is uploaded.

Splits the bill

Bill X-ray: re-reading, cache writes and output, then long chats, cold resumes and model switches.

Counts down the cache

Cache clock: a countdown on every paused chat, 5 minutes or 1 hour, before the next turn writes it all again.

Flags unused tools

Tools tab: tokens each skill and plugin costs a turn, its calls in this chat, and one switch for the unused ones.

Trims a chat by topic

Context blocks: remove a block, topic or keyword, and a new chat carries the rest word for word. Its quality is not measured.

Our token-saving ideas, tested on real tasks

We set each pass rule before the run, and most ideas failed. We do not claim that Fleet will save you money. The X-ray shows your numbers; this table shows ours.

Tests 4 levers with fair coins

Each new chat draws on or off for every lever, and Fleet compares both groups on your work, with guardrails.

{"chat":"c42","hc":0,"sm":1,"cap":1}

Keeps rules out of compaction

Pinned rules such as "Ask before you push" live in the system prompt, which compaction never rewrites.

• always ask before you push

Sends totals only when you click

The Share tab opens a GitHub issue with totals only. You read it first and press send.

No project names. No chat text.

Others found the same trap. In a 2,908-run study of Claude Code, one compression setup cut tool output by 38.4% and raised the billed cost by 6.8%. Prompt caching is the reason.

4 failed3 mixed or not proven1 passed, narrowly scoped
IdeaTestResultVerdictWhat we did
Fused search and edit tools12 paired tasks with strict checksAbout 11% more expensiveFailedRemoved the claim
All read-only helpers on Haiku14 tasks, blind judge, control arm98% cheaper. Score 7.0 vs 8.07 of 10. Worst task lost 4FailedNot a default
All read-only helpers on SonnetSame 14 tasks65% cheaper. Score 6.71 vs 8.07. Worst task lost 5FailedNot a default
Haiku for mechanical lookups and log searches only14 new tasks, scored by computerSame answers as Opus on all 14, 97% cheaper. These tasks were easy for every modelPassed, narrowly scopedSaver's lookup helper uses it. Not a claim about your bill
Compact at 100K, no tools10 runs on 6 real chats. A worst-case memory quizRecall 3.9 vs 6.3 of 8FailedOff by default
Compact on resume at 200K5 real chats, real next message, control armFacts 7.4 vs 8.0 of 8. Next-step plans 1.2 vs 2.0 of 2Facts passed, plans failedOn, with the result shown in Settings. You can turn it off
Pinned rules: standing rules kept in the system promptPaired test, 4 rules checked by script, rule set first. Chat rules vs pinned rules after compaction, plus a no-compaction controlRules said in the chat were broken 9% of the time after compaction. Pinned: 0%. Control: 6%No measured benefitOptional. We do not claim it improves rule-following
Compact only after a pause, not at a fixed sizeReplay of 30 days of one person's chats. No model calls, quality not measuredReplay: about 23% cheaper for chats and no compaction in the middle of a tool loop. A plain 200K cap: about 18% and 15 mid-loop compactionsReplay only, not provenOpt-in "Smart compact". Fleet logs each setting change so the effect can be checked on real use

Small samples, on one person's work, at API list prices. The last row replays past chats: it shows where a rule would have saved money, not what it does to quality.

Free for 3 panes, Pro for $24 once

Fleet does not resell tokens, so you pay us once and your vendor as usual.

Free

$0

  • 14 days with every feature, no card
  • Then 3 panes in total, with every feature
  • Bill X-ray with no limit

Pro, one time

$24launch price, $39 after

  • Up to 16 panes per screen, unlimited screens
  • Works on 3 of your Macs
  • Every 0.x update. 30-day refund
Buy Fleet Pro

Local tax may apply at checkout. Payments go through Dodo Payments.

What Fleet reads and what it sends

Fleet runs the official claude and codex commands on your Mac. It does not modify them or read your login tokens.

Your usage
It stays between you and your vendor. Fleet is not in that path.
Your transcripts
The X-ray, the Tools tab and the cache clock read them here and upload 0 bytes. No analytics.
Two network calls
A license check after you buy, and appcast.xml for updates. Little Snitch shows both.

Install and questions

READYFleet 1.0.1 · notarized by Apple

Download Fleet 1.0.1

Fleet is signed, notarized by Apple and passes 278 automated checks. It needs macOS 13 or later on Apple Silicon. Open the DMG and drag Fleet to Applications.

Download the DMG, 5.3 MB
Or install from the terminal

It downloads the app, checks its SHA-256 and puts it in Applications. Read the script first.

curl -fsSL https://fleetapp.in/install.sh | sh

Or take the DMG from the latest release.

Does it work with my Claude plan?

Yes. It runs Claude Code in your own account, the same as a terminal. You sign in through Claude Code, not through Fleet.

Why not tmux or terminal tabs?

If those work for you, keep them. Fleet is for when you run enough agents that finding the one waiting for you has become the job.

Will it save me tokens?

We do not promise that. See the test table. The X-ray shows where your usage goes, and what you change is up to you.

What does Fleet not do yet?
  • It runs on Mac only, Apple Silicon only for now. No Linux or Windows.
  • No remote access from your phone or another machine.
  • If you quit Fleet, its agents stop. Chats resume by id; a task in progress does not.
  • Codex panes show DONE, not WAITING, and have no per-pane cost.
  • Updates install with one click, but Fleet never installs one on its own.
  • It is a beta from one developer. Expect bugs and tell me about them.
Is this made by Anthropic or OpenAI?

No. Fleet is an independent app and is not affiliated with or endorsed by either company. Claude Code and Codex belong to their owners.