What is Verne?
Verne is an open-source desktop IDE for your coding agents. It wraps Claude Code, Codex, or any CLI agent in a real workspace: terminals, a Monaco editor, an embedded browser, git, and per-workspace notes. The agent does its work in the terminal; you follow along everywhere else.
Run several agents and projects side by side, and Verne shows which ones are working, blocked, or idle without you reading each terminal. Everything runs on your machine: Verne talks to the agents you’ve already installed, and your code and prompts never leave your computer.
Who Verne is for
- Developers who already drive CLI agents like
claudeorcodexand want more than a bare terminal around them. - Anyone running multiple agents or projects at once who loses track of which one needs attention.
- People who want to review an agent’s diff before it commits, in a real editor with git built in.
- Teams who want agents to drive a browser and read notes over MCP, instead of working blind.
When to use Verne
Verne is a good fit when you:
- Run more than one agent and want their status at a glance.
- Need agents that keep running after you close the window, ready to reattach when you reopen Verne.
- Want to isolate risky work in a git worktree without leaving the app.
- Would rather keep the editor, git review, and browser next to the agent than in three other windows.
Running a single quick command? A plain terminal is fine. Verne helps most once you’re tracking more than one thing at a time.
What you get
- The Agents panel: every agent across every workspace, with live working / blocked / idle status.
- Persistent sessions: a background daemon keeps terminals and agents alive, ready to reattach.
- A built-in code editor: Monaco (the editor behind VS Code), with LSP, themes, and familiar keybindings.
- Source control & review: see every changed file, read a line-by-line diff, and commit in place.
- An embedded browser: preview your work, then let agents click, type, and navigate it over MCP.
- Agent-readable notes: per-workspace notes your agents read over MCP, so context lives next to the code.
Next steps
- Installation: download Verne and get the
verneCLI on your PATH. - Your first session: open a folder, launch an agent, and see it working.
- Supported agents: which CLI agents Verne detects and integrates with.