Personal Claude assistant designed for secure, containerized execution, targeting users who prioritize understanding and customizing their software. It offers a lightweight, AI-native alternative to complex frameworks, providing a transparent, secure, and adaptable personal AI assistant.

How It Works

NanoClaw employs a single Node.js process, executing agents within isolated Linux containers (Apple Container on macOS, Docker on Linux/macOS). This isolation, coupled with explicit filesystem mounting, enhances security. Customization and setup are managed conversationally via "Claude Code," an AI assistant, eschewing traditional configuration files. This approach prioritizes code clarity and direct user control over abstract configurations.

Quick Start & Requirements

• Primary install / run command: Navigate to the cloned repository directory and execute /setup.
• Non-default prerequisites and dependencies: Node.js 20+, Claude Code, Apple Container (macOS) or Docker (macOS/Linux).
• Estimated setup time or resource footprint: Setup is guided interactively by Claude Code.
• Links: https://github.com/gavrielc/nanoclaw.git

Highlighted Details

• WhatsApp I/O integration for messaging.
• Isolated group contexts with per-group memory (CLAUDE.md) and sandboxed containers.
• Scheduled tasks for recurring AI-driven actions.
• Web access capabilities for information retrieval.
• Customization via natural language prompts to Claude Code, directly modifying the codebase.
• AI-native interaction for setup, debugging, and management.

Maintenance & Community

The project focuses on individual customization via forking. While specific contributors like @dotsetgreg are noted for skills, explicit community links (Discord/Slack) or sponsorship details are not detailed in the README.

Licensing & Compatibility

MIT License. The project aims for legitimate use with the Claude Agent SDK, avoiding terms of service ambiguities. Primary compatibility is for macOS and Linux environments.

Limitations & Caveats

Primarily engineered for a single user's specific needs, not as a general-purpose framework. Customization necessitates direct code modification, guided by Claude Code. While Docker is supported on macOS, Apple Container is highlighted as optimized for Apple Silicon. Windows support is noted as a potential future "skill."