OpenClaw

Ecosystem

OpenClaw has inspired an ecosystem of variant frameworks across multiple languages and platforms. From 500-line minimal builds to ESP32 embedded agents, there is a flavor for every use case.

NanoClaw

7K
TypeScript 500 lines, container isolation

Minimal, auditable reimplementation of OpenClaw in just 500 lines. Uses Apple containers for security isolation and is built on Anthropic's Agents SDK.

Best for

Security, Auditability, Learning

View on GitHub

Nanobot

17.8K
Python 45MB RAM, 0.8s cold start

Ultra-lightweight Python alternative with a 99% smaller codebase. Two-command setup, MCP-native, and supports OpenRouter, Anthropic, OpenAI, DeepSeek, and Gemini.

Best for

Lightweight, Multi-platform, Resource-constrained

View on GitHub

PicoClaw

5K
Go <10MB RAM, single binary

Targets $10 RISC-V boards with 400x faster startup. 95% agent-generated core with no external dependencies. Ships as a single binary.

Best for

IoT, Embedded, Edge computing

View on GitHub

ClawGo

Go Official Go implementation

The official Go implementation of OpenClaw, maintained as part of the core project. Designed for Go ecosystems and production server environments.

Best for

Go ecosystems, Production servers, Microservices

View on GitHub

IronClaw

Rust WASM sandbox

Maintained by NEAR AI, IronClaw runs every untrusted tool in an isolated WebAssembly container with explicit capability-based permissions.

Best for

Security-critical, WASM, Zero-trust environments

View on GitHub

MicroClaw

129
Rust Channel-agnostic core

Inspired by NanoClaw and rewritten in Rust. Channel-agnostic core that supports Telegram, Discord, and Web. Handles up to 100 iterations per request.

Best for

Custom platforms, Library usage, Embedding

View on GitHub

MimiClaw

C ESP32, $5 chip

Runs on ESP32-S3 with 16MB flash and 8MB PSRAM at just 0.5W. Implements a ReAct agent loop in C with OTA firmware updates and dual-core support.

Best for

Bare-metal IoT, ESP32, Hardware projects

TinyClaw

23
TypeScript Self-improving, learning

An OpenClaw variant that learns from conversations over time. Multi-provider support for Anthropic, OpenAI, and Ollama with automatic provider fallback.

Best for

Personal companions, Learning agents, Research

View on GitHub

At a glance

Compare variant frameworks by language, size, and target use case.

Name Language Stars Key Differentiator Best For
NanoClaw TypeScript 7K 500 lines, container isolation Security, Auditability, Learning
Nanobot Python 17.8K 45MB RAM, 0.8s cold start Lightweight, Multi-platform, Resource-constrained
PicoClaw Go 5K <10MB RAM, single binary IoT, Embedded, Edge computing
ClawGo Go -- Official Go implementation Go ecosystems, Production servers, Microservices
IronClaw Rust -- WASM sandbox Security-critical, WASM, Zero-trust environments
MicroClaw Rust 129 Channel-agnostic core Custom platforms, Library usage, Embedding
MimiClaw C -- ESP32, $5 chip Bare-metal IoT, ESP32, Hardware projects
TinyClaw TypeScript 23 Self-improving, learning Personal companions, Learning agents, Research

Build your own variant

OpenClaw's architecture is open and well-documented. Whether you want to rewrite the core in a new language, target a specific platform, or experiment with novel agent patterns, the community is here to support you.

Want the original?

Deploy the official OpenClaw in minutes with npm, Docker, or one-click cloud platforms.