HolyClaude
One-command AI development workstation: Claude Code, web UI, 8 AI CLIs, headless browser, and 50+ dev tools — containerized and ready.
Evidence shows the container runs as non-root by default and provides PUID/PGID mapping, but the compose template includes broad capabilities like SYS_ADMIN, SYS_PTRACE, and seccomp=unconfined, and there is no user confirmation mechanism. Data flow transparency is partially addressed: README explains credential storage and direct provider connections, but internal data flows are not detailed. Sensitive data handling includes scripts to protect credential file permissions, but no full encryption or key management. Dependency security includes SBOM and vulnerability scanning, but no specific vulnerability reports. External effects: container binds to localhost by default, but SSH and port mapping options are provided. Rollback: version tags and upgrade instructions exist, but no automatic rollback mechanism. Source attribution: clear author and license, but publisher is unverified.
Self-consistency: version numbers in README, compose files, and test scripts are consistent, but no complete architecture documentation. Dependency availability: Docker images and dependency locking exist, but no offline installation option. Failure messages: test scripts have detailed error output, but user documentation troubleshooting is incomplete.
Audience and scenarios: README clearly targets developers and provides multiple use cases. Capability boundaries: lists included tools and features, but no explicit limitations. Trigger precision: provides environment variables and configuration options, but no detailed trigger conditions. Environment fit: supports multiple platforms and architectures, but no Kubernetes deployment.
Information architecture: README is well-structured with table of contents and detailed sections. Install notes: provides quick start and full configuration templates. Naming stability: version and tag naming is consistent. Examples and FAQ: provides examples and common issues, but no complete FAQ. Known limitations: has a known issues section, but not detailed. License: MIT license is clear. Versioning and changelog: has CHANGELOG and version tags. Maintenance responsibility: has contribution guidelines and support channels, but no explicit maintainer.
Output usability: provides web UI and CLI, output formats are clear. Marginal value: provides pre-configured AI development environment, saving setup time. Cost benefit: free and open source, but requires user's own AI subscription or API keys.
Claim traceability: version and feature claims in README are consistent with checks in test scripts. Cross-source corroboration: information on Docker Hub and GitHub is consistent. Fact-inference separation: README distinguishes facts and inferences, but no detailed verification methods.
- The container includes broad capabilities like SYS_ADMIN, SYS_PTRACE, and seccomp=unconfined by default; consider restricting these in untrusted environments.
- The publisher is unverified; review image provenance and integrity before deployment.
- Dependency security scan results are not public; run your own vulnerability scanning.
What does this agent do, and when should you use it?
HolyClaude is a pre-configured Docker container that delivers a complete AI coding environment out of the box. It bundles the Claude Code CLI with a web UI via CloudCLI, Chromium headless browser with Playwright, eight AI CLIs (including Gemini, Codex, Cursor), and over 50 development tools (Node.js, Python packages, system utilities). The container uses s6-overlay for process supervision, ensuring auto-restart and graceful shutdown. Data and workspace are persisted via bind mounts, and it supports both AMD64 and ARM64 architectures with full and slim image variants. HolyClaude works with existing Anthropic subscriptions for authentication without extra configuration.
HolyClaude sets up a development environment inside a Docker container. On startup, it launches the CloudCLI web UI on port 3001, providing a browser interface to Claude Code. The environment includes pre-installed Chromium, Xvfb, and Playwright for browser automation and screenshots. It supports multiple AI providers through eight bundled CLIs: claude, gemini, codex, cursor, task-master, junie, opencode, and pi. Configuration for these providers is done via environment variables like GEMINI_API_KEY, OPENAI_API_KEY, and CURSOR_API_KEY. Notifications via Apprise (Discord, Telegram, etc.) can be enabled with simple environment variables. The container uses s6-overlay for process management, auto-restarting crashed services. Persistent configuration and workspace are stored in ./data/claude and ./workspace respectively, surviving container rebuilds.
- A developer who wants to skip the 1-2 hour manual setup of Claude Code, Chromium, and Playwright, and start with a single docker compose up.
- Self-hosting an AI coding workstation on a Synology NAS or Raspberry Pi (ARM64) with secure remote access via Tailscale or Cloudflare Tunnel.
- A team that wants a uniform AI coding environment (Claude Code, Gemini, Codex, etc.) across all members using the same Docker image.
- A QA engineer needing ready-to-use browser automation and screenshot capabilities with Playwright and Chromium, already configured.
- An AI enthusiast who wants to compare outputs from eight different AI CLIs (Claude, Gemini, Codex, etc.) without installing each separately.
- A developer working in an environment with limited disk space who opts for the slim image and lets Claude install additional packages on demand.
What are this agent's strengths and limitations?
- Saves 1-2 hours of manual setup with a 30-second docker compose up that pre-configures tools, browser, and permissions.
- Includes over 50 dev tools: Node.js 26, Python 3 with popular packages (pandas, numpy, playwright), system tools (ripgrep, git, gh), and more.
- Supports eight AI CLIs in one container, eliminating separate installs and configuration for each provider.
- Uses s6-overlay for process supervision with auto-restart, crash recovery, and graceful shutdown.
- Persistent bind mounts (./data/claude, ./workspace) survive container rebuilds, preserving Claude config, auth, and code.
- Slim image (about 2.4GB) available for disk-constrained environments, with on-demand package installation via Claude.
- Full image is heavy (about 4.1GB compressed), requiring significant disk and bandwidth.
- Primary dependency on Claude Code; other AI CLIs require their own subscriptions or API keys, adding provider lock-in.
- Default Docker config requires extra capabilities (SYS_ADMIN, SYS_PTRACE, seccomp=unconfined) and 2GB shared memory, which may raise security concerns.
- CloudCLI web UI bound to localhost; remote access requires setting up Tailscale or Cloudflare Tunnel, adding complexity.
- CloudCLI account is not persisted by default, so it must be recreated after container rebuild (quick but takes ~10 seconds).
- Documentation strongly advises against exposing the CloudCLI port to the public internet, citing high security risk.
How do you install or deploy this agent?
Requires Docker and Docker Compose. Create a folder, then create a docker-compose.yaml file. You can use the 'Quick Start' template (minimal) or the 'Full Configuration' template (with all options) from the README. Then run 'docker compose up -d'. By default, the web UI is exposed on 127.0.0.1:3001.
How do you use this agent?
After starting, open http://localhost:3001 in your browser. Create a CloudCLI account (takes about 10 seconds), then sign in with your Anthropic account (supports OAuth from Max/Pro plan or an API key). You can now use Claude Code through the web UI, or access the command line inside the container to run 'claude', 'gemini', 'codex', etc. For other AI CLIs, set the required environment variables (e.g., GEMINI_API_KEY, OPENAI_API_KEY).
How does this agent compare with similar options?
Compared to other Dockerized AI coding environments, HolyClaude offers a more comprehensive pre-configured setup, including eight AI CLIs, headless browser, and s6-overlay process management. Many similar projects might only provide a Docker image for Claude Code without the web UI, browser automation, or multi-provider support.