Documentation
Everything you need to run Verde.
Verde is a native desktop workspace for coding agents — Codex, Claude Code, OpenCode, Cursor, and Amp side by side in one tiling window, as native chat panes or each agent's own TUI. These docs cover installation, provider setup, the tiled workspace, keybinds, the scriptable CLI, configuration, and troubleshooting.
These docs are crawlable as plain markdown. Pull llms.txt for a compact index, llms-full.txt for everything in one file, or grab any page as raw markdown at /docs/<slug>.md (e.g. /docs/quickstart.md).
GET /llms.txtGet started
Install Verde, authenticate a provider, and run your first chat thread.
Workspace
The tiled workspace — panes, splits, focus, resize, the terminal dock, and keybinds.
Chat, models & runs
Provider readiness, searchable model selection, run controls, follow-ups, approvals, and transcript tool-call groups.
Design Mode
Select elements or regions in the embedded browser and route visual feedback to a Verde chat or terminal agent.
Panes & tiling
How Verde's tiled workspace is laid out — splits, focus, resize, zoom, the sidebar, the terminal dock, and the embedded browser pane.
Keybinds
Every default keybind in Verde, plus how to remap or disable each binding in verde.json.
Reference
The full verde CLI surface, configuration files, themes, logs, and troubleshooting.
CLI reference
The verde CLI surface — updates, offline state, live browser and chat control, persistent terminal sessions, integrations, completion, and exit codes.
Configuration & state
verde.json settings, update and transcript preferences, verde.yml stacks, themes, state files, and runtime logs.
Troubleshooting
Provider readiness, browser and Design Mode support, updates, source-build errors, and runtime logs.
Verde is built with Zig, SDL3, and Ghostty. Source on GitHub.