The desktop runner. Native to your machine.
Two agents. One runner.
Station runs both Claude Code and Codex, on macOS, Linux and Windows, so every engineer works in the agent they are fastest in, for writing code and for reviewing it. Station launches each agent with the model, effort and speed you chose for it in Build.
- Claude Code and Codex side by side on the same board, on all three platforms.
- Per-agent model settings for coding and for review, applied on every launch.
- Each running session shows its runtime, so you always know what is doing the work.
Every task in its own workspace. Approved work, landed.
Each task gets its own isolated copy of the repository, so parallel work never collides and a review always reads the exact commit it is judging. A session that writes code is sandboxed: on macOS and Linux it cannot read your credential stores, Station's own data or the checkouts of unrelated projects, and on Windows it cannot reach Station's data. When a person approves the work, Station can land it for you.
- Every task runs in its own isolated workspace. Your own checkout stays yours.
- Sandboxed sessions on macOS, Linux and Windows. Agents can use podman and docker without your registry credentials.
- Approved work lands on the task's target branch, as a merge or as a pull request, exactly as it was approved.
- Merges are recognised wherever they happen, so the board knows what shipped.
Choose how each Station takes part.
One setting per project decides how this Station takes part: Build starts sessions here, a person drives the work by hand while Build still tracks and reviews it, or this machine sits the project out. Every engineer's machine contributes exactly as much as it should.
- Per project, per Station: automatic sessions, manual only, or off.
- Manual work still flows through the board, the review and the record.
- Cursor and GitHub Copilot get configured against the same checkout.
A desktop app, native to your machine.
Station ships as a native desktop app: small, fast and light on your machine. Universal macOS build (Intel + Apple Silicon), .msi and .exe installers for Windows, and a native Linux build. Tray icon, menu controls, and a real native presence.
- Native on every platform. Small download, low overhead.
- macOS universal binary; .msi and .exe installers for Windows; native build for Linux.
- Tray icon with Show / Hide, Pause All, Resume All, Refresh MCP, Settings, Quit.
Scan locally. No extra setup.
The architecture scanner runs in-process inside Station. No separate tool to install, no extra toolchain. The scan walks your repo, detects surfaces, and builds a fixed-schema envelope that uploads to Server for the team. Your source never leaves the machine.
- Built in to Station. Nothing extra to install.
- Fixed envelope schema with `repo_sha`, `scan_id`, and timestamps for reproducible bookkeeping.
- Regenerate on demand. Server stores the result so the team sees the same thing.
Watch the agent work. From any browser.
Agent output streams from the workstation to the dashboard in real time. The full Claude Code or Codex session (edits, test runs, decisions) visible to anyone with access to the project. Streams are resilient across reconnects, sleep/wake, and deploy restarts.
- Stdout, file edits, and tool calls stream live, not via polling.
- Resilient transport survives laptop sleep, network blips, and server restarts.
- Non-developers (PMs, engineering managers) can watch without an operator install.
⌘K everything.
The command palette opens on ⌘K and covers running tasks, switching projects, and jumping between views. A tabbed Settings page keeps general, project, and about configuration in one place.
- Fuzzy search across tasks, projects, and recent actions, keyboard-first.
- Tabbed Settings (General / Projects / About). Nothing buried two levels deep.
- Sidebar nav: Home, Agents, Projects, Logs.
Runs unattended. Doesn't hang on you.
Headless agents run to completion unattended, recover from sleep/wake cleanly, and keep their session logs for as long as you choose. Long-running CI usage works as well as interactive operator usage.
- Headless agents run unattended, with no prompts waiting on an operator.
- Reconnects after sleep / wake or network blips happen automatically.
- Session log retention in days, set by you; structured logs with live level switching.
Station is the GUI runner. One of three.
Put Station on every machine that ships code.
Station is the agent runtime; Server is the control plane it reports to, managed by us or hosted by you. Tell us about your team and we'll walk through Station rollout with your real environment, including how it fits your managed-software setup.