Portfolio control tower
Organizations → workspaces → projects → tasks on one Kanban board. Run dozens of apps without losing the thread.
Put an AI crew at the helm of your whole portfolio.
Command Fleet is the local-first command center for builders running many projects at once. Dispatch any task to Claude Code, Codex, or Gemini — then review the diff and merge.
Orchestrates
Claude Code
Codex
Gemini
Code never leaves your machine
Isolated git worktree per run
Keys in your OS vault
Bring your own agent subscriptions
The problem
Every AI coding tool watches one repo. You ship a dozen.
Founders, indie studios, and agencies don't have one app — they have a portfolio. Switching IDEs, juggling terminals, and losing track of what each agent did doesn't scale. Command Fleet is the layer above the agents: one place to run, watch, and ship them all.
Why Command Fleet
Your whole portfolio, under command
Six things no single-repo assistant gives you.
Any agent, per task
Claude Code, Codex, or Gemini — choose the right one for each job and switch whenever you like. No lock-in.
Isolated worktrees
Every run gets its own branch and git worktree, so parallel agents never clobber each other — or your working tree.
Review then merge
An in-app diff shows exactly what changed with +/- coloring. When it looks right, merge with one click — or discard cleanly.
Autonomous PM
Describe an app in a sentence. The workspace manager scaffolds it, plans the tasks, builds, reviews, merges, and deploys.
Local-first & private
Projects, data, and API keys stay on your computer. Nothing is uploaded — keys live in your OS credential vault.
The landscape
There's an open quadrant. We're standing in it.
Most tools manage a single repo, or live in the cloud. Command Fleet is the only one built for a whole portfolio, autonomously, on your own machine.
From a sentence to shipped
An autonomous PM that actually delegates
Tell the workspace manager “build me a waitlist app on Firebase.” It scaffolds the project from a stack pack, breaks the work into an ordered task graph, and dispatches each task to an agent in its own worktree.
- Dependency-aware planning — tasks run in the right order, in parallel where they can
- Self-review loop — failed work is retried automatically, then escalated to you
- Auto-merge — finished branches integrate so the next task builds on real progress
- Preview & deploy built in — six stack packs from Firebase to Fly
How it works
A workflow that feels like delegating
You stay the reviewer. The agents do the typing.
1
Add a project
Point Command Fleet at a folder and describe what you want in plain language.
2
Dispatch to an agent
Pick Claude Code, Codex, or Gemini and hit run. An isolated worktree spins up automatically.
3
Watch it work
Streaming output lands in the task history in real time, cancellable anytime.
4
Review & merge
Inspect every change in-app, then merge the branch — or discard it cleanly.
How we compare
The fleet vs. the field
| Capability | Command Fleet | Single-agent GUIs | Cloud platforms |
|---|---|---|---|
| Manage many projects at once | Yes | No | Limited |
| Agent-agnostic (Claude / Codex / Gemini) | Yes | Usually one | Locked |
| Runs fully local / private | Yes | Yes | No |
| Autonomous plan → build → deploy | Yes | No | Yes |
| Bring your own subscriptions | Yes | Yes | Pay per platform |
See the full breakdown on the comparison page.
3
AI agents, one app
∞
Projects & tasks
6
Deploy stack packs
100%
Local & private
From the deck
Built for people who run a lot of things
“I went from babysitting one terminal to running five projects from a single board. The review-before-merge step is what makes it safe.”
Indie founder · 7 live apps
“Local-first was the dealbreaker for our agency. Client code never touches a cloud we don't control, and we still get autonomous builds.”
Studio lead · dev agency
“Switching agents per task is underrated. Cheap model for boilerplate, the smart one for the gnarly refactor — all in one place.”
Solo developer · SaaS portfolio
Questions
Frequently asked
Is my code sent to the cloud?
No. Command Fleet runs entirely on your machine. Your projects, tasks, and data never leave your computer — only the agent CLIs you choose talk to their own providers, exactly as they would in your terminal.
Which AI agents are supported?
Claude Code, Codex, and Gemini — chosen per task, switchable anytime. You bring your own subscriptions or API keys, stored in your OS credential vault.
What do I need to run it?
Windows and the agent CLI you want to use on your PATH. Command Fleet handles the worktrees, streaming, diffing, merging, and deploys around it.
Do I pay for the AI usage too?
Yes — you use your existing Claude, Codex, or Gemini plan. Command Fleet is the orchestration layer on top, so you're never double-charged for model usage.
Put a fleet at the helm today.
Start your 7-day free trial — no credit card. Your portfolio is about to move a lot faster.