Skip to content
Get packs

NanoClaw

A small, container-isolated AI assistant that connects to chat apps and runs on Anthropic's Agent SDK.

GitHub Download v2026.10.0
Stars
31.1k
Forks
12.7k
Licence
MIT
Language
TypeScript
Last update
4 days ago
Type
AI model / agent

What it is

NanoClaw is an open-source AI assistant you run on your own machine. It connects to messaging apps such as WhatsApp, Telegram, Slack, Discord and others, and each agent runs in its own Linux container so it can only see files you explicitly mount. The README describes it as a smaller alternative to OpenClaw, designed to be read and modified as code rather than configured through many settings.

  • Chat with a named agent over Slack, Telegram, Discord or WhatsApp
  • Schedule recurring jobs that the agent runs on its own
  • Give each agent its own CLAUDE.md instructions and memory
  • Use a local CLI channel to talk to an agent without a chat app

Great for

  • Developers who want a small codebase they can read and change
  • People who want an assistant on chat apps with container-level isolation
  • Users comfortable working with Claude Code to customize their fork
  • Anyone who wants separate agents with separate memory per channel

Not ideal for

  • Non-technical users who want a finished app with no setup, since install runs shell scripts and may require Docker and Node
  • Users on native Windows, since the README lists macOS, Linux and WSL2 for containers
  • People who expect a monitoring dashboard or debugging UI, which the README says is intentionally absent
Step-by-step guide: how to use NanoClawInstall, first run, examples, tips and FAQs
NanoClaw alternativesNanoClaw vs OpenClawNanoClaw vs OpenHandsEdict vs NanoClaw

How to get it

  1. 1

    Clone and build from source

    You'll need: Git · Docker (installed by the script if missing) · Node.js and pnpm (installed by the script if missing)

    Clone the repository

    git clone https://github.com/nanocoai/nanoclaw.git nanoclaw-v2

    Enter the directory

    cd nanoclaw-v2

    Run the setup script

    bash nanoclaw.sh
  2. 2

    Run it

    • Run bash nanoclaw.sh and follow the prompts
    • Let the script register your Anthropic credential and build the agent container
    • Pair your first channel (Slack, Telegram, Discord, WhatsApp, iMessage or local CLI)
    • Message your named agent to confirm it responds
  3. 3

    Get the most out of it

    • Add channels on demand with /add-<channel> skills so you only install what you need
    • Choose a channel isolation model per channel via /manage-channels
    • Set up scheduled tasks with script gates to avoid unnecessary agent runs
    • Use agent templates from the templates folder via ncl groups create --template

Downloads

Source code (.zip)main branch · straight from GitHubRelease v2026.10.0Release notes and sourceOfficial website & docsnanoclaw.dev

Strengths

  • Agents run in containers and only see explicitly mounted files, per the README
  • The codebase is described as small, with one process and a handful of files
  • Supports many channels and can use Claude, Codex, OpenCode, Cursor or local Ollama models

Trade-offs

  • Relies on Claude Code and Anthropic's Agent SDK for its default workflow, so it assumes that stack
  • Setup and customization depend on Claude Code, and the README says there is no monitoring UI
  • Some features, such as the credential gateway and prebuilt image, involve a third-party gateway or optional account

How it compares

ProjectStarsTypeLevelUpdatedBest for
nanoclaw (this)31.1kAI model / agentintermediate4 days agoDevelopers who want a small codebase they can read and change
openclaw394kAppintermediate4 days agoPeople who want an AI assistant that runs on their own machine rather than a hosted service
OpenHands92.5kAppintermediate4 days agoDevelopers who want coding agents running on a server while their laptop is off
paperclip110kAppadvanced4 days agoDevelopers already running several AI coding agents who want one place to track them
marketingskills57.9kLearning resourcebeginner6 days agoTechnical marketers who already use AI coding agents
awesome-agent-skills36.8kCurated listbeginner4 days agoDevelopers who use Claude Code, Codex, Cursor, or Gemini CLI and want ready-made skills

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.