DeckSpace is the command deck for agentic coding: run every CLI agent you use across parallel panes, each isolated on its own git worktree, with a board, a swarm, diff review, memory and voice — so ideas move from prompt to product without leaving one window.
DeckSpace$ implement checkout fix
→ editing api/checkout.ts
✓ wrote 3 files
$ review diff #worktree-2
→ inspecting 42 changes
notes captured
$ npm test
✓ unit suite
✓ typecheck
› worktree ready
› awaiting task
checkout-bug-#214
DeckSpace gives agentic coding a home base: the board feeds tasks into the deck, terminal panes run the work, each agent moves in parallel on its own git worktree, and diff review stays close enough to keep you in control.
A real terminal grid — xterm + native PTYs, not a toy sandbox. Launch one agent or fill the deck and watch every pane move at once.
Every parallel task runs on its own branch + worktree, so agents never trip over each other's files. Review and merge each one when it's ready.
The work has one center of gravity. Pull a card from the board, spin up the crew, and decide what ships from the diff — the human stays in the loop.
DeckSpace launches the CLI agents you already have installed and authenticated — mix and match models across panes. No new accounts, no lock-in. Bring your own key; DeckSpace never sees or stores it.
A living knowledge graph lives next to your code as plain markdown. Every agent in the deck — builders, reviewers, voice — reads and writes the same local hub through MCP. Decisions stick, context compounds, and the next session starts where the last one ended.
Notes are files you own — commit, version and back them up like any other asset.
Backlinks connect today's bug to the note you wrote three weeks ago.
Start from a Kanban card, a repo, or a rough idea. The session becomes a focused deck instead of a pile of tabs.
Terminal panes, file tree, editor and agent launch points come up together — one place for the work.
Spin up one agent or decompose a goal into a parallel crew on isolated worktrees, and watch status change live.
Jump between code, shell output and the diff. A verification gate runs tests, lint and a secret scan before you merge.
Focused spaces for the parts of building that usually scatter across windows, tabs and half-finished prompts.
Keep real shells visible, organized and tied to the task they're running — with a destructive-op guard on by request.
Decompose a goal, launch builders and reviewers on isolated worktrees, and merge each result behind a review gate.
Inspect the diff, see the verification verdict, and decide when the work is ready — with checkpoints to undo anything.
The point isn't to hide complexity — it's to keep the moving parts visible enough that you can steer them.
Real processes and projects, no toy sandbox.
Parallel agents never collide on files.
Tests, lint & secret scan before merge.
On-device Whisper dictation with ⌘⇧D.
Checkout bug pulled from the board.
Repo, shells, editor and worktrees open together.
Builder implements while the reviewer watches the diff.
Review notes, test output and changed files are ready.
DeckSpace is a native desktop app (Tauri + Rust) — small, fast and offline-capable. The terminal, editor, board and memory all run locally. The only network calls are the ones your own agent CLIs make to your own model provider.
Use the CLIs you already pay for. DeckSpace never meters or marks up your model usage, and never adds its own charge.
Confine an agent's writes to the repo, reroute destructive commands to a recoverable trash, and snapshot before every action.
Block cloud-metadata endpoints and SSRF. Telemetry is off by default — nothing about your code goes anywhere.
An agent development environment — a native desktop "command deck" for running AI coding agents. Instead of juggling terminals, tabs and prompts, you get one workspace: a grid of agent panes, a Kanban board, a swarm, diff review, a memory graph and voice, all wired together.
Cursor is an IDE; Warp is a terminal. DeckSpace is the room around them — built to orchestrate multiple agents in parallel, each isolated on its own git worktree, with a board feeding the work and a review gate at the end. It runs the CLI agents you already use rather than replacing your editor.
Eight CLI agents today: Claude Code, Codex, Gemini, Aider, OpenCode, Cursor Agent, Droid and Copilot CLI. DeckSpace detects whichever ones you have on your PATH and lets you mix models across panes.
You authenticate the agent CLIs once, the way you already do — DeckSpace never sees or stores your keys, and never adds its own charge. Prefer not to manage a key? The Pro+ plan bundles the models for you.
No. The terminal, editor, board and memory all run locally and offline. The only outbound calls are the ones your own agent CLIs make to your own model provider. Telemetry is off by default.
macOS, Windows and Linux. It's a native Tauri + Rust app — not an Electron shell — so it stays small, fast and offline-capable.
The core is free forever and bring-your-own-key. Pro adds the heavy lifting at a flat fee — no metered credits, no surprise overages. The 14-day trial unlocks everything.
Early access is rolling out now. Tell us where to send the build for macOS, Windows or Linux — we'll be in touch fast.
Thanks — we'll email you the DeckSpace build and your access details shortly.