Someone finally containerized Claude Code without the grief

What it does

HolyClaude is a fat Docker container that pre-installs the official Anthropic Claude Code CLI, a browser-based UI called CloudCLI, a headless Chromium stack with Playwright, and roughly fifty development tools. It also tosses in several competing AI CLIs—Codex, Gemini, Cursor, TaskMaster, Junie, and OpenCode—so you can switch models without switching terminals. Everything runs under s6-overlay process supervision with bind-mounted persistence for credentials.

The interesting bit

The author isn’t just bundling software; they’re fixing the specific, tedious failures that happen when you run Claude Code inside Docker—shared memory limits, root-owned WORKDIR hanging the installer, UID mismatches causing permission denials, and SQLite locking on NAS SMB mounts. It is essentially a bug diary turned into infrastructure.

Key highlights

• Runs the real Claude Code CLI, not a proxy; authenticates via existing Max/Pro subscription or API key through the web UI
• Multi-arch AMD64 and ARM64 builds, including support for Apple Silicon, Raspberry Pi, and Synology/QNAP NAS
• Pre-configured headless browser stack (Chromium + Xvfb + Playwright) ready for screenshots and testing
• Includes s6-overlay for auto-restart and graceful shutdown of internal services
• Credentials stay in a local bind-mounted volume; the container does not intercept or proxy billing

Caveats

• Kubernetes support is marked “coming soon” with a Helm chart planned but not yet available
• The slim variant installs extras on-demand, which implies the latest image is hefty; exact image sizes aren’t stated in the README

Verdict

Grab this if you want a turnkey, browser-accessible AI coding environment on a home server or VPS without manually wrestling Docker permissions. Skip it if you prefer to compose your own minimal images or don’t need a web UI wrapper around terminal-based tools.