NanoClaw — Lightweight Personal AI Assistant
A minimal alternative to OpenClaw that runs Claude agents in isolated containers. Connects to WhatsApp, Telegram, Slack, Discord, and Gmail with just 500 lines of TypeScript.
Review-first install path
This asset needs a review step. The copied prompt tells the agent to dry-run, show the writes, then proceed only after confirmation.
npx -y tokrepo@latest install b399753b-8f80-4a07-81ac-e985d166d5dd --target codexDry-run first, confirm the writes, then run this command.
What it is
NanoClaw is a lightweight personal AI assistant that runs Claude agents in isolated Docker containers and connects them to messaging platforms. It supports WhatsApp, Telegram, Slack, Discord, and Gmail as communication channels. The entire codebase is roughly 500 lines of TypeScript, making it easy to understand and modify.
Individuals who want a personal AI assistant accessible across their messaging apps and developers looking for a minimal, self-hosted alternative to heavier AI assistant frameworks will find NanoClaw practical. It is designed as a simpler alternative to OpenClaw.
How it saves time or tokens
Setting up an AI assistant that works across multiple messaging platforms typically requires building separate integrations for each service. NanoClaw provides pre-built connectors for five platforms in a single project. Running in Docker containers provides isolation and reproducibility. The minimal codebase means you can customize behavior without navigating a large framework.
How to use
- Clone the repository:
git clone https://github.com/qwibitai/nanoclaw.git
cd nanoclaw
- Run Claude Code in the project directory:
claude
- Inside the Claude session, run the setup skill:
/setup
- Claude Code reads the project CLAUDE.md and walks you through configuring messaging platform credentials and environment variables.
Example
# Clone and set up
git clone https://github.com/qwibitai/nanoclaw.git
cd nanoclaw
# Configure environment
cp .env.example .env
# Edit .env with your API keys:
# ANTHROPIC_API_KEY=your-key
# TELEGRAM_BOT_TOKEN=your-token
# SLACK_BOT_TOKEN=xoxb-your-token
# Build and run
docker compose up -d
# The assistant is now reachable on your configured platforms
Related on TokRepo
- AI Agent Tools -- explore AI agent frameworks and assistants
- Self-Hosted Tools -- discover self-hosted AI tools you can run on your own server
Common pitfalls
- Each messaging platform requires its own API credentials. Set up bot tokens for Telegram, app tokens for Slack, and OAuth for Gmail before starting the service.
- Docker is required for container isolation. NanoClaw does not run natively without containerization.
- The assistant uses Claude as its LLM backend, so you need a valid Anthropic API key. Token costs depend on conversation volume across all connected platforms.
Frequently Asked Questions
NanoClaw supports WhatsApp, Telegram, Slack, Discord, and Gmail. Each platform requires its own API credentials configured in the environment file. The assistant maintains separate conversation contexts per platform and per user.
NanoClaw is a minimal alternative to OpenClaw, built with roughly 500 lines of TypeScript compared to OpenClaw larger codebase. It focuses on simplicity and ease of modification while providing the core functionality of running Claude agents with multi-platform messaging integration.
Yes. NanoClaw runs Claude agents in isolated Docker containers for security and reproducibility. Docker and Docker Compose are prerequisites. The container setup handles dependency management and environment isolation.
Yes. The codebase is intentionally small (around 500 lines of TypeScript) to make customization straightforward. You can modify the system prompt, add custom tools, change conversation handling logic, or add new messaging platform connectors.
NanoClaw uses Anthropic Claude as its LLM backend. You need a valid ANTHROPIC_API_KEY in your environment configuration. The assistant sends messages from your connected platforms to Claude and returns the responses through the same channel.
Citations (3)
- NanoClaw GitHub— Minimal AI assistant running Claude in containers with multi-platform messaging
- Anthropic Claude API— Claude API for AI agent capabilities
- Docker Documentation— Docker containerization for isolated agent execution
Related on TokRepo
Source & Thanks
Discussion
Related Assets
Cloud Hypervisor — Lightweight Virtual Machine Monitor in Rust
Cloud Hypervisor is a KVM-based virtual machine monitor written in Rust for modern cloud workloads. It provides a minimal attack surface, fast boot times, and support for CPU/memory/device hotplug with a focus on security and simplicity.
pbi-cli — Power BI Skills for Claude Code
pbi-cli is a Python CLI that installs Claude Code skills for Power BI models and PBIR reports. Get started with pipx + `skills install`.
Turf.js — Advanced Geospatial Analysis for JavaScript
A modular geospatial analysis engine written in JavaScript and TypeScript for browsers and Node.js.
Fluentd — Unified Logging Layer for Cloud-Native Infrastructure
Fluentd is a CNCF-graduated open-source data collector that unifies log collection and routing. With 1000+ plugins, it connects any source to any destination — the standard log layer for Kubernetes alongside Fluent Bit.