NanoClaw
Your personal AI agent. Secure. Lightweight. Yours.
The official website for NanoClaw, the secure personal AI agent. NanoClaw runs securely in containers, built to be understood and customized for your own needs.
Why NanoClaw?
NanoClaw delivers the same core functionality in a codebase you can actually understand.
 | OpenClaw | |
|---|---|---|
| Source files | 15 | 3,680 |
| Lines of code | ~3,900 | 434,453 |
| Dependencies | <10 | 70 |
| Config files | 0 | 53 |
| Time to understand | 8 minutes | 1–2 weeks |
| Security model | OS container isolation | Application-level checks |
| Architecture | Single process + isolated containers | Single process, shared memory |
What It Supports
Everything NanoClaw supports out of the box, and nothing you don't need.
Multi-channel messaging
WhatsApp, Telegram, Discord, Slack, Microsoft Teams, iMessage, Matrix, Google Chat, Webex, Linear, GitHub, WeChat, and email. Installed on demand with /add-<channel> skills. Run one or many at the same time.
Container isolation
Agents are sandboxed in Docker (macOS, Linux, WSL2), with optional Docker Sandboxes micro-VM isolation or Apple Container as a macOS-native opt-in.
Flexible isolation V2
Connect each channel to its own agent for full privacy, share one agent across many channels for unified memory, or fold multiple channels into a single shared session. Pick per channel.
Per-agent workspace
Each agent group has its own CLAUDE.md, its own memory, its own container, and only the mounts you allow. Nothing crosses the boundary unless you wire it to.
Scheduled tasks
Recurring jobs that run Claude and message you back. Morning briefings, weekly reviews, and more.
Skills over features
Trunk ships the registry and infrastructure. Install channels and providers with /add-<name>; the skill copies exactly the module(s) you need into your fork.
AI-native, hybrid by design
A fast scripted install for the happy path; Claude Code takes over when a step needs judgment. No dashboards or debug UI beyond setup — describe the problem in chat.
Credential security
Agents never hold raw API keys. Outbound requests route through OneCLI's Agent Vault, which injects credentials at request time and enforces per-agent policies and rate limits.
Architecture
A single Node host orchestrates per-session agent containers. Two SQLite files per session, each with exactly one writer — no cross-mount contention, no IPC, no stdin piping.
Single host process
One Node host routes inbound via the entity model (user → messaging group → agent group → session), writes to inbound.db, and wakes the container. No microservices, no message brokers.
Per-agent-group containers
Each agent group runs in its own container with its own CLAUDE.md, memory, skills, and only the mounts you allow. Nothing crosses the boundary unless you wire it to.
Credential isolation
Outbound HTTPS routes through OneCLI's Agent Vault, which injects credentials at request time and enforces per-agent policies and rate limits. Agents never hold raw API keys.
Self-registering extensions
Channels and alternative providers self-register at startup. Trunk ships the registry and the Chat SDK bridge; adapters themselves are skill-installed per fork.
Key Files
src/index.ts — Entry point — DB init, channel adapters, delivery polls, sweep src/container-runner.ts — Spawns per-agent-group containers, OneCLI credential injection src/router.ts — Inbound routing: messaging group → agent group → session → inbound.db src/delivery.ts — Polls outbound.db, delivers via adapter, handles system actions src/db/ — Central DB — users, roles, agent groups, messaging groups, wiring, migrations src/host-sweep.ts — 60s sweep — stale detection, due-message wake, recurrence Philosophy
The principles that shape every NanoClaw decision.
Small enough to understand
One process, a few source files, no microservices. If you want to understand the full codebase, ask Claude Code to walk you through it.
Secure by isolation
Agents run in Linux containers and can only see what's explicitly mounted. Bash access is safe because commands run inside the container, not on your host.
Built for the individual user
NanoClaw isn't a monolithic framework; it's software that fits each user's exact needs. Instead of becoming bloatware, it's designed to be bespoke. Fork it and have Claude Code shape it to you.
AI-native, hybrid by design
The install and onboarding flow is scripted, fast, and deterministic. When a step needs judgment — a failed install, a guided decision, a customization — control hands off to Claude Code seamlessly.
Skills over features
Trunk ships the registry and infrastructure, not channel adapters or alternative providers. Channels live on a channels branch, providers on providers. Run /add-telegram, /add-opencode, etc. and the skill copies exactly what you need into your fork.
Best harness, best model
Natively runs Claude Code via the official Claude Agent SDK. Other providers are drop-in options: /add-codex (OpenAI), /add-opencode (OpenRouter, Google, DeepSeek, and more), /add-ollama-provider (local open-weight models). Provider is configurable per agent group.
Get Started with NanoClaw in 3 Lines
Clone NanoClaw, enter the directory, and run the install script.
$ git clone https://github.com/qwibitai/nanoclaw.git nanoclaw-v2
$ cd nanoclaw-v2
$ bash nanoclaw.sh nanoclaw.sh walks you from a fresh machine to a named agent you can message. It installs Node, pnpm, and Docker if missing, registers your Anthropic credential with OneCLI, builds the agent container, and pairs your first channel. If a step fails, Claude Code is invoked automatically to diagnose and resume.
Requirements
In the News
