Our take
Activeregular commits and releases- Merges more pull requests than 85% of the projects we track
- More contributors than 77% of the projects we track
NanoClaw is an actively maintained, MIT-licensed option for users who value understandable code and container boundaries, with a setup and customization model that assumes technical involvement.
Good fit if
- You want a self-hosted assistant connected to messaging channels, with agents isolated in containers.
- You are comfortable with Node.js, Docker, and adapting a codebase or using Claude Code during setup and customization.
- You want an actively maintained project: it has recent commits and releases, over 100 contributors, and prompt issue and pull-request handling, without measured contributor concentration flagged as a risk.
Look elsewhere if
- You need a configuration-first product or a built-in monitoring and debugging dashboard. The README explicitly favors code customization and chat-based debugging.
- You cannot run Docker-based agent containers or do not want to manage a self-hosted installation.
- You need a turnkey assistant that does not depend on an Anthropic-based setup path or optional skills for alternative providers and channels.
All health signals
| Last commit | 2026-10-06 (2 days ago) |
|---|---|
| Commits, last 90 days | 100+ |
| Releases, last 12 months | 10 (latest v2026.10.0-rc.2, 2026-10-05, pre-release) |
| Contributors | 100+ (top contributor: 52% of commits) |
| Issues closed, last 90 days | 9+ (typically closed in 3 days) |
| Pull requests merged, last 90 days | 76 (typically merged in 2 days) |
| Project age | 8 months |
Checked on 2026-10-08 with the GitHub API.
Overview
NanoClaw is a self-hosted AI assistant that routes messages from supported channels to agents running in separate Linux containers. Its goal is to offer messaging, memory, and scheduled tasks without requiring users to trust a large application with unrestricted access to their machine.
The project is designed for people willing to customize their own checkout: its philosophy favors code changes and installable skills over configuration files and a feature-rich central framework. The host coordinates messages and sessions, while each agent runs with only the files explicitly mounted for it.
Key Features
- Routes messages between supported channels and isolated agent sessions.
- Runs agents in Docker containers with per-agent workspaces, memory, and explicitly allowed mounts.
- Supports scheduled tasks, including optional script gates to avoid waking an agent when there is no work.
- Keeps credentials outside agent containers and routes outbound requests through a credential gateway.
- Offers per-agent provider choices through skills, including Claude, OpenAI Codex, OpenCode, and Ollama options described in the README.
- Provides agent templates that bundle instructions, tools, and skills without secrets.
- Supports different channel-to-agent arrangements, from separate agents to shared sessions.
Use Cases
- An individual can message a personal assistant through a channel they already use and give it access only to selected folders.
- A household or small team can keep separate agent memories and containers while using shared chat rooms where appropriate.
- A user can schedule recurring research or briefing tasks and receive the results through a messaging channel.
- A technically comfortable user can fork the project and adapt its behavior directly in code rather than maintaining extensive configuration.
What you need
Detected in the repository
- Node.js >=22 (from package.json)
- Automated checks on GitHub Actions
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
git clone https://github.com/nanocoai/nanoclaw.git nanoclaw-v2
cd nanoclaw-v2
bash nanoclaw.sh
See the README for setup details, channel and provider skills, and migration instructions.
Alternatives
- mercury-agent: Mercury offers CLI, Telegram, and web access with permission prompts and persistent memory, while NanoClaw emphasizes isolated containers and messaging channels.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| nanoclaw | TypeScript | MIT | 30.9k | Active |
| mercury-agent | TypeScript | MIT | 3.2k | New |
Considerations
- This is a fast-moving project, not a long-established release: it is about 250 days old, with frequent commits and releases. The latest listed release is a prerelease, so teams should review changes before relying on an upgrade.
- Setup expects Node.js 22 or newer and uses Docker for agent isolation. The README says the setup script can install Node and Docker when missing, but running containerized agents still requires Docker support.
- The README describes Claude Code as part of setup recovery and customization, and says channels and alternative providers are installed on demand through skills rather than all shipping in the core.
- The project emphasizes code-level customization instead of configuration files and does not provide a monitoring dashboard or debugging UI.
- Agents are isolated by mounts, but adopting the project still means reviewing its code, chosen channel integrations, and credential-gateway setup for your environment.
- The MIT license permits broad reuse, subject to its license terms.
Found this useful?
Share it with someone who would like nanoclaw.