iloveyuko21-star/moltbot-sandbox-100

OpenClaw on Cloudflare Workers: A Personal AI Assistant in the Cloud

The moltbot-sandbox-100 project represents an experimental implementation of OpenClaw (formerly known as Moltbot and Clawdbot) running entirely within Cloudflare's Sandbox environment. This proof-of-concept demonstrates how a fully-featured personal AI assistant can be deployed on Cloudflare Workers without traditional self-hosting infrastructure, leveraging Cloudflare's serverless container technology to provide an always-on, managed deployment.

It's important to note that this is explicitly labeled as experimental and unsupported, meaning it may break without notice. However, for developers interested in cloudflare-workers or exploring ai-assistant deployment patterns, it offers a fascinating architectural approach.

Key Features and Capabilities

OpenClaw itself is a comprehensive personal AI assistant built on a gateway architecture that connects to multiple chat platforms. The core feature set includes:

• Control UI - A web-based chat interface accessible directly through the gateway, providing a centralized point of interaction
• Multi-channel support - Native integrations with Telegram, Discord, and Slack, allowing you to interact with your assistant from your preferred platform
• Device pairing - Secure direct message authentication requiring explicit approval, ensuring only authorized devices can access the assistant
• Persistent conversations - Chat history and context maintenance across sessions when R2 storage is enabled
• Agent runtime - An extensible AI capabilities framework with workspace and skills support

The Cloudflare deployment adds another layer of functionality through integration with cloudflare-ai-gateway for API routing and analytics, Cloudflare Access for authentication, Browser Rendering for navigation capabilities, and optional R2 Storage for persistence across container restarts.

Technical Stack and Architecture

Built primarily in TypeScript, this project leverages several Cloudflare platform features to create a cohesive deployment:

• Cloudflare Sandbox Containers - Runs on a standard-1 instance (1/2 vCPU, 4 GiB memory, 8 GB disk)
• Cloudflare Workers - Requires the paid plan at $5/month to access Sandbox containers
• Anthropic Claude API - Powers the AI capabilities, either through direct API access or via Cloudflare AI Gateway's unified billing
• Cloudflare Access - Provides authentication and security for the admin interface
• Optional R2 Storage - Enables data persistence for paired devices and conversation history

The repository includes comprehensive configuration files including Docker support, Vite and Vitest configurations for development and testing, and Wrangler configuration for Cloudflare deployment. The project structure suggests a well-organized codebase with dedicated directories for skills, end-to-end tests, and browser automation capabilities.

Installation and Setup Process

Setting up moltbot-sandbox-100 requires several steps and careful attention to security configuration. The deployment process begins with installing dependencies via npm, followed by configuring essential secrets including your Anthropic API key and a generated gateway token for remote access. The gateway token is particularly important as you cannot access the Control UI without it.

A critical post-deployment requirement is configuring Cloudflare Access to protect the admin UI at /_admin/. This involves enabling Cloudflare Access on your workers.dev domain, setting the appropriate CF_ACCESS_TEAM_DOMAIN and CF_ACCESS_AUD secrets, and configuring authorized users through the Zero Trust Dashboard. Without this configuration, the admin UI where device pairing occurs remains inaccessible.

The documentation provides clear guidance on cost estimation, noting that a 24/7 running container would cost approximately $34.50/month including the Workers plan fee. However, it recommends configuring the SANDBOX_SLEEP_AFTER parameter to reduce costs significantly—a container running only 4 hours per day might cost just $5-6/month in compute plus the plan fee.

Community and Development Status

As a newly published project (last updated February 7, 2026), moltbot-sandbox-100 currently shows 0 stars, 0 forks, and 0 open issues on GitHub. The repository has a single commit and no active community engagement yet. However, it's released under the permissive Apache-2.0 license, which encourages both commercial and non-commercial use and contributions.

The project includes CONTRIBUTING.md and AGENTS.md documentation files, suggesting the maintainer welcomes community participation and has considered extensibility through the agent framework. For developers interested in ai-chatbot development or serverless-deployment, this represents an interesting starting point despite its early stage.

Comparison with Alternatives

Compared to traditional self-hosted AI assistants or platforms like Rasa, Botpress, or custom GPT implementations, moltbot-sandbox-100 offers a unique serverless approach. Traditional solutions require managing servers, handling scaling, and maintaining uptime, while this Cloudflare-based approach provides managed infrastructure with built-in security through Cloudflare Access.

The trade-off is cost predictability and the experimental nature of the implementation. At $34.50/month for continuous operation, it's more expensive than basic VPS hosting but includes enterprise-grade features like DDoS protection, global edge distribution, and integrated authentication. The sleep functionality can bring costs down to competitive levels for intermittent use.

For developers already invested in the Cloudflare ecosystem or those seeking a deployment option that doesn't require infrastructure management, this project offers a compelling alternative to traditional hosting approaches for personal AI assistants.