CloudCLI – Web UI for Claude Code, Cursor CLI, and Codex

TL;DR

TL;DR: CloudCLI wraps Claude Code, Cursor CLI, Codex, and Gemini CLI in persistent cloud containers with a web UI, file explorer, Git integration, and cross-device access. Self-host free or use the hosted plan from €7/month.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: cloudcli.ai — verified OG image and feature list
• Source repository: github.com/siteboon/claudecodeui — read README end-to-end
• License: AGPL-3.0 (verified from LICENSE file)
• GitHub metadata: 12,522 stars, 1,700+ forks, topics include anthropic, claude-code, claude-code-ui, cursor, gui, mobile-first, react
• HN launch thread: news.ycombinator.com/item?id=45427482

What Is CloudCLI?

CloudCLI (also called Claude Code UI) is an open-source web interface for running AI coding agents — specifically Claude Code, Cursor CLI, Codex, and Gemini CLI — in persistent cloud development environments.

The core problem it solves: Claude Code’s Remote Control feature depends on your laptop staying awake and connected. Close the lid or lose internet for about 10 minutes and the session dies. CloudCLI runs the agent in a cloud container that persists regardless of your local machine state. You can start work on your phone, switch to VS Code over SSH, and pick up the same session later.

The project has 12,522 GitHub stars and is actively maintained. The hosted version at cloudcli.ai starts at €7/month; the full open-source version is free to self-host.

Setup Workflow

Option 1 – CloudCLI Cloud (Hosted)

The fastest path. No local setup required.

1. Sign up at cloudcli.ai
2. Choose a plan (Hobby from €7/month)
3. Start a new environment — Claude Code, Cursor CLI, Codex, or Gemini CLI is pre-installed in each container
4. Access via browser, mobile app, or SSH

Option 2 – Self-Hosted (Open Source)

Requires Node.js v22+.

# Instant try with npx (no install)
npx @cloudcli-ai/cloudcli

# Or install globally
npm install -g @cloudcli-ai/cloudcli
cloudcli

Open http://localhost:3001 after starting. Your existing local sessions are discovered automatically on first launch.

For remote server deployment, Docker support, and PM2 setup, see the official documentation.

Docker Sandboxes (Experimental)

Agents can also run in hypervisor-isolated Docker sandboxes. Requires the sbx CLI:

sbx run --agent claude-code

Features

• Responsive Web UI — desktop, tablet, and mobile layouts. Built as a PWA for installability.
• Multi-Agent Support — Claude Code, Cursor CLI, Codex, Gemini CLI, and OpenCode in the same workspace.
• Integrated Terminal — direct shell access to the agent’s CLI without leaving the browser.
• File Explorer — interactive file tree with syntax highlighting and live editing.
• Git Explorer — stage, commit, switch branches, view diffs from the UI.
• Browser Use — open browser sessions for web research and agent-driven browser automation.
• Session Management — resume prior conversations, manage multiple sessions, track history.
• Plugin System — extend CloudCLI with custom tabs, backend services, and integrations. Starter template at github.com/cloudcli-ai/cloudcli-plugin-starter.
• REST API — trigger sessions, retrieve results programmatically. Used by n8n, Linear, Jira integrations.
• Cross-Device Access — start on mobile, continue in VS Code over SSH, all with the same persistent session.
• Model Compatibility — works with any Claude or GPT model family. Full model list available at runtime via GET /api/providers/:provider/models.

Practical Evaluation Checklist

• [ ] Pulled the GitHub README and verified all commands
• [ ] Confirmed AGPL-3.0 license from LICENSE file
• [ ] Verified 12,522 stars via GitHub API
• [ ] Checked OG image (1200×630, 36.6KB) from cloudcli.ai
• [ ] Confirmed Node.js v22+ requirement for self-hosted
• [ ] Verified npx @cloudcli-ai/cloudcli as the install command (not npm install of a differently-named package)
• [ ] Cross-device session persistence claim matches README

Security Notes

• Each user gets an isolated container — code is not shared between users
• API keys and credentials stored encrypted within the environment
• Self-hosting option available for teams with strict data sovereignty requirements
• All connections are encrypted

FAQ

Q: Do I need my own AI subscription? A: Yes. CloudCLI provides the cloud environment and tooling, not the AI model itself. You bring your own API key for Claude Code, Cursor, Codex, or Gemini.

Q: How is this different from Claude Code Remote Control? A: Remote Control requires your laptop to stay awake and connected. Cloud sessions die when the lid closes or network drops. CloudCLI runs in a persistent cloud container, so closing your laptop has no effect.

Q: How is this different from GitHub Codespaces? A: Codespaces is a general-purpose cloud dev environment. CloudCLI is purpose-built for AI coding agents, comes with Claude Code, Cursor CLI, Codex, and Gemini CLI pre-installed, includes a mobile-optimized agent UI, supports GitLab and Bitbucket (not just GitHub), and starts at a flat €7/month vs. Codespaces’ per-minute billing.

Q: Can I self-host it? A: Yes. The claudecodeui repo is fully open source under AGPL-3. Run it locally with npx @cloudcli-ai/cloudcli, deploy on a VPS, or use a cloud provider of your choice.

Q: Can teams share configurations? A: Yes. CloudCLI supports shared MCP servers, CLAUDE.md, .cursorrules, environment variables, and tool configurations across projects. Each developer still has an isolated container.

Conclusion

CloudCLI solves the laptop-dependency problem that makes Claude Code Remote Control impractical for mobile or multi-device workflows. By running the agent in a persistent cloud container, it decouples the coding session from any single machine — start on your phone, continue in VS Code, nothing gets lost. The open-source version is genuinely free, the hosted tier is competitive with Codespaces, and the mobile-first design is well-executed. If you use Claude Code, Cursor CLI, or Codex regularly and want a cross-device workflow, it is worth trying.

Repo: github.com/siteboon/claudecodeui Docs: cloudcli.ai/docs