nanocoai/nanoclaw

What it solves

NanoClaw is a lightweight AI assistant framework designed for users who want a secure, customizable, and private way to run AI agents across multiple messaging platforms. It addresses the security risks of running complex AI software with broad system access by isolating agents in their own Linux containers, ensuring they only access explicitly mounted directories.

How it works

NanoClaw uses a single Node.js host process that acts as a router. When a message arrives from a messaging app (like Telegram or Discord), the host routes it to the specific agent group's session and writes it to an inbound SQLite database. A sandboxed Docker container running the Claude Agent SDK (via Bun) polls this database, processes the request, and writes the response to an outbound SQLite database, which the host then delivers back to the user.

Who it’s for

It is built for individual users who prefer a minimal, "AI-native" codebase that they can easily understand, fork, and customize using AI tools like Claude Code, rather than relying on a large, monolithic framework with complex configuration files.

Highlights

• Container Isolation: Agents run in Docker containers to prevent host system access.
• Multi-channel Support: On-demand installation of adapters for WhatsApp, Telegram, Discord, Slack, and more.
• Credential Security: API keys are managed via OneCLI's Agent Vault, so agents never hold raw keys.
• Bespoke Customization: Designed to be modified directly in the code via AI prompts rather than through config files.
• Scheduled Tasks: Ability to set up recurring jobs with optional script gates.
• Flexible Model Providers: Native support for Claude, with options for OpenAI, Google, DeepSeek, and local models via Ollama.