The shared development environment
for humans and agents.
atrium is a persistent, programmable workspace for real CLI coding agents. Launch them, resume them, steer them, and let them coordinate inside a room that survives crashes, reboots, and reopenings.
Most AI tools give agents a task. atrium gives humans and agents a shared environment.
Between AI IDEs
and blind agent orchestration.
IDEs are built around a human in a file. Orchestration systems are built around agents running unattended.
atrium lives in the gap between them: a shared development environment where humans can steer, inspect, and collaborate with real coding agents over time.
AI IDEs
Built around a human in a file
atrium
A shared environment where humans steer, inspect, and collaborate with real coding agents over time
Unattended orchestration
Built around agents running without human oversight
The room survives
atrium restores panes, layout, scrollback, working directories, branch and worktree context, and compatible agent sessions after a quit, crash, reboot, or reopen.
Humans and agents work in the same room
Terminals, files, editors, git, search, and worktrees live alongside active coding agents in one workspace instead of being scattered across separate tools.
Agents can coordinate inside the workspace
Compatible agents can discover panes, inspect workspace context, read visible output, and send messages across panes with scope and visibility controls.
A programmable environment under the hood
atrium is not just a shell with tabs. Its internal protocol and state model make the workspace addressable. Agents are not merely running inside the app — they are operating within a shared environment the app can restore, inspect, and coordinate.
Human
atrium workspace
Persistent · Programmable · Shared
Protocol · State · MCP
Agents
atrium:// protocol
URI-based dispatch for pane, workspace, command, hook, and MCP operations
MCP tools
Agents list panes, read output, inspect context, and write to other panes
State-backed workspace model
Every namespace addressable, restorable, and observable via the state registry
Shared memory and ambient awareness
The next layer of atrium expands the shared room into a shared cognitive environment.
Shared Memory
Agents retain and retrieve workspace context across sessions instead of constantly rebuilding it from scratch.
Ambient Awareness
Agents become aware of relevant activity elsewhere in the workspace, reducing blind collisions and improving coordination.
Everything around the work is already there
Worktree-aware file tree
Follows the focused terminal's branch. Git status inline. Drag files into panes.
Monaco code editor
The same engine behind VS Code, for opening and editing files in-place.
Markdown editor
Rich text editing with comments, table of contents, and Mermaid diagrams.
Built-in source control
Stage, diff, commit, push, and drill into commits without leaving the workspace.
Repo search
Search file contents and filenames with regex, live streaming results.
Native workspace controls
Split panes, drag-to-rearrange, tabs with layout diagrams, global zoom, themes.
Why atrium instead of just using
an IDE or another terminal
Strong for file-centric assistance, weak for persistent multi-agent workspace operations.
Strong for command execution, weak for persistence and agent semantics.
Strong for unattended automation, weak for shared human-agent workspace control.
A persistent, programmable environment where humans steer real coding agents inside the same room.
Built to stay open all day
Open the room once.
Come back to it anytime.
Built for developers working with real coding agents today, with a shared environment that gets more capable as the agents inside it do.
Claude Code and Codex support built in, with a growing adapter model underneath. macOS today. Windows and Linux coming.