DeckSpacemacOS
Apple Silicon and Intel. DMG installer or Mac zip. macOS 12 or newer.
Mac installation guide →Your coding agents,on one deck.
Run the agent CLIs you already use in real terminal panes, side by side — start a task, follow the run, review the diff and keep control of the work.
Pro · $24/month · Paid upfront · 7-day money-back guarantee · or $228/year
Interactive demo · sample workspace
Full screen ↗
The real DeckSpace interface on a simulated workspace — agent runs are replayed, nothing executes, and every room and pane is the actual app. Loads only when you press play. Prefer it big? Open full screen ↗
Workspace
Run them side by side.
Every pane is a real terminal running the CLI you chose. Give each agent its own task, watch them work in parallel, and step in the moment one needs you.
Panes
Panes are real terminals
Native PTYs running any CLI on your PATH — up to 16 at once. Rename a workspace and it keeps its identity; the layout is remembered per workspace.
Team
One worktree per agent
Hand a goal to a team and each worker gets its own git worktree. Board → Team → Review, so parallel work never collides and nothing merges unreviewed.
In the window
Check it without leaving
Diff review, a Browser room for the running app, the DeckTest engine and region screenshots live next to the panes — the evidence stays with the work.
Voice & Remote
Speak it. Steer it from your phone.
Hold a key and talk to the deck. Step away, and the free iPhone app shows every pane live — type into one, send a key, answer the question that is blocking a run.
DeckSpace Remote · free
Atlas & dictation
Talk to the deck
Hold a key, speak, release. Dictate a prompt into any pane or ask Atlas to open a room, spawn an agent or run a mission.
DeckSpace Remote
Free on the App Store
Pair by QR on your own network. Watch every pane, type into one, approve a plan or answer a question from your phone — and revoke access any time. How the phone remote works →
Routines
Scheduled runs
Put a task on a schedule and let it run while the app is open. Unattended runs inherit the sandbox and fail closed. What routines can and cannot do →
Verification
See it work before you merge.
An agent that says “done” is a claim. The deck keeps the evidence next to the diff so you can check it in seconds.
✓42 tests passedpane 1 · claude
✓browser flow · sign-up → dashboardDeckTest
✓screenshot attached · regionshots/…png
?one question waiting for youneeds you
⇥review gate · nothing merges unrevieweddiff
DeckTest
Runs the checks, shows the result
Point it at the running app and it drives a real browser through the flow, then reports what broke with the screenshot to prove it. The engine is bundled with every DeckSpace plan; the standalone app is $29 on its own ↗ or included with Deck Complete.
Screenshots
Show the agent what you see
Capture a region on a hotkey and drop the image path straight into the prompt. Nothing is uploaded anywhere.
Review
Nothing merges unreviewed
A/B runs pin their base and stay reviewable after any outcome. Agents can request a helper or a pane; they cannot install, authenticate or launch one themselves.
Inside the deck
The rest of the toolkit, wired to the same agents.
One memory, one board, one set of controls — every tool sees the same context.
Iterate
Hand it a mission and it builds, checks and fixes in a loop. You set the goal and the stopping rule. Permissions for unattended runs →
Team
Decompose one goal into a coordinated team, each on its own worktree, merging behind the review gate. Swarm and worktrees →
Memory
A local knowledge graph in plain markdown, served to your agents over MCP. Context compounds session to session. Memory →
DeckLink · MCP
Self-hostable MCP servers hand your memory graph and your board to any agent tool that speaks MCP. Plugins →
Git, transport and GitHub connection states are reported separately, GitHub connect and clone run under your own account, and Code Search respects ignore rules and only promotes a complete index. Read the changelog →
Bring your own agents
The CLIs you already have. Your accounts, your keys.
DeckSpace launches the agent CLIs installed and authenticated on your machine, and detects whichever are on your PATH. Model usage is billed by your provider, with no DeckSpace markup. Which CLIs, and how to register your own →
Claude CodeCodexAiderAntigravityOpenCodeCursor AgentDroidCopilot CLIGrok CLIAmpQwen CodeGemini (legacy CLI)
Claude Accounts $14 · ADD-ON
Run several Claude accounts side by side — one desktop app per account, each logged in once. Included with Deck Complete.
Local & private
Your workspace stays local. You choose the providers.
bring your own key
No usage markup
Use the CLIs you already pay for. DeckSpace adds no markup to model usage — you pay your provider directly. DeckSpace itself is a flat subscription, never per-token.
on your machine
Files, panes and memory stay local
Project files, terminal sessions, Board and Memory are stored on your device. Cloud agents send only the context needed for the requests you make to the provider you chose.
contained
Sandbox & guard
Use the operating-system fence and shell guards to limit agent writes. Unattended runs inherit the sandbox. Check the protection status before launching; support varies by platform.
Pricing
One subscription. The whole deck.
New subscriptions are paid upfront. 7-day money-back guarantee; cancel future renewals anytime. Flat fees, bring your own key, no metering. Compare plans and the one-time apps →
Pro
$24 /mo
billed $24 monthly
The full command deck. No metered credits, no surprise bills.
- Paid upfront
- Up to 16 panes in parallel
- Team orchestration on isolated worktrees
- Background + scheduled agents · loops
- Atlas voice control · free iPhone Remote
- Repo-aware autocomplete + code search MCP
- A/B model compare · bundled DeckTest engine
Most popular
Deck Complete
$39 /mo
billed $39 monthly
DeckSpace, DeckVoice, DeckTest and Claude Accounts — one subscription.
- Everything in Pro, plus
- DeckVoice and DeckTest licences, plus Claude Accounts, while subscribed
- Priority support
- One invoice for the whole deck
Team
$69 /seat/mo
2–50 seats
Deck Complete for every seat, managed in one place.
- Everything in Complete, plus
- 2–50 seats on one team licence · 2 devices per seat
- Priority support
Or own a tool outright — DeckVoice $19 · DeckTest $29 · Claude Accounts $14. Complete licences deactivate if you cancel; owned licences never do.
Install
Your project. Your desktop.
Pick your platform on the personal download page you receive after checkout. Your licence key activates the app.
Windows
Windows 10 and 11, 64-bit. Setup installer, or MSI for managed installs.
Windows installation guide →Linux
AppImage, Debian/Ubuntu .deb or Fedora .rpm. Check the guide for compatible desktops.
Linux installation guide →iPhone Free
DeckSpace Remote pairs with the desktop app over your own network. Watch, type, approve and answer from your phone.
Download on the App Store →Bought already? Manage billing or re-send your licence key.
Questions
Frequently asked.
What is DeckSpace?
A native desktop command deck for running AI coding agents in parallel — a grid of real terminal panes, a board, teams on isolated worktrees, diff review, a local memory graph, loops and voice. Around it sit DeckVoice, DeckTest and Claude Accounts, all sharing the same agents.
Do I need an account?
No account. Your licence key and the email you paid with are how billing and support recognise you. Billing, key re-send and tickets live on the Help page.
Which agents are supported?
Claude Code, Codex, Aider, Antigravity, OpenCode, Cursor Agent, Droid, Copilot CLI, Grok CLI, Amp, Qwen Code and a legacy Gemini entry. You can also register any other CLI yourself; DeckSpace detects whichever are on your PATH.
Do I pay per credit?
No. DeckSpace is bring-your-own-key — you authenticate the agent CLIs once and DeckSpace never meters, marks up or brokers your model usage. Pro is $24/mo (or $228/yr). Deck Complete is $39/mo (or $372/yr) and adds DeckVoice, DeckTest and Claude Accounts while you subscribe. Team is $69 per seat per month for 2 to 50 seats. New subscriptions are paid upfront. 7-day money-back guarantee.
Does my code leave my machine?
Workspace files and history stay on your computer. Cloud agents and enabled AI features send relevant code, prompts or audio to your selected providers. Local transcription runs on your device. Licensing, downloads and connected services also use the network.
Can I pick up where I left off?
DeckSpace restores your workspace layout and reopens confirmed Claude and Codex conversations using their saved IDs. If a conversation cannot be confirmed, a visible recovery choice keeps you in control. Read about conversation recovery.
How does the iPhone Remote work?
Install the free DeckSpace Remote app, scan the QR code the desktop app shows, and the two pair over your own network. Remote sessions close their terminal streams the moment access is revoked or narrowed. What the phone can and cannot do.
What platforms does it run on?
macOS 12+, Windows 10/11 (64-bit), and Linux (64-bit). Choose your installer on the download page and check the install guide for signing and desktop requirements. Standalone apps have their own platform requirements.
Community
Build in the open with us.
Release notes, workflows people actually use, and a place to ask. One email when something ships — nothing else. See the community page →
YouTube · @DeckSpacedevFrom info@deckspace.dev · unsubscribe any time · never shared
Get DeckSpace
Give your agents a command deck.
For Mac, Windows and Linux, with a free iPhone Remote. Subscribe and pay upfront, then download the installer for your platform.
$24/mo, paid upfront · cancel anytime · bring your own key · local workspace · macOS, Windows and Linux