Skip to main content

The Pulse of
OpenClaw Clones

Track the Claws. Pick the Sharpest. AI-powered insights for the most ambitious open-source models and their clones.

Explore Clones
742.1k+ Total Stars
Decision Entry Points

Best OpenClaw Alternatives by Use Case

Pick the lane first, then inspect the shortlist. Each card gives you a fast starting point before you dive into full clone profiles.

See Full Grid
Most Secure

Start here if shell risk keeps you up at night.

Top Pick

Prioritize isolation, telemetry discipline, and safer defaults before raw breadth.

IronClaw
Rust 11.9k
Local First

Use this shortlist when data sovereignty matters most.

Top Pick

These projects skew toward on-device inference, tighter data boundaries, or lower cloud dependence.

SafeClaw
Python 257
Zero API Cost

The shortlist for builders trying to dodge permanent token bills.

Top Pick

Great starting point for offline-first workflows, local model usage, or deterministic stacks.

SafeClaw
Python 257
For Teams

Pick from the projects that look more collaborative than personal.

Top Pick

Bias toward multi-user coordination, workplace channels, orchestration, and enterprise posture.

Hermes-agent
Python 110.7k
Edge / Lightweight

For constrained hardware, fast boot, and minimal runtime overhead.

Top Pick

Best for builders targeting tiny VPS instances, ARM boards, embedded gear, or ultra-lean deployments.

MimiClaw
C 5.2k
Best OpenClaw Replacement

The first shortlist to open when you want fewer compromises than OpenClaw.

Top Pick

Weighted toward credibility, momentum, safety posture, and how complete the alternative feels today.

Hermes-agent
Python 110.7k
ClawVerse Hub

Claw Clones

Comparing active forks and variants. Use the decision entry points above for a fast shortlist, then use the grid below to inspect the full field.

Jump to Use Cases
43
Active Nodes Tracked
Decision Filter
All Models

Full ecosystem view

Moltis

Rust

A Rust-native AI agent framework that prioritizes security through sandboxed execution and local key storage. Delivers a single-binary experience with voice, memory, and multi-platform integrations—all without the Node.js runtime baggage.

Pulse
78%
Release
Apr 22, 2026
2,594
7
35%

GitClaw

TypeScript

GitClaw reimagines AI agents as fully version-controlled git repositories where identity, rules, memory, and tools are all trackable files. It's a meta twist on the OpenClaw concept — fork an agent, branch its personality, and `git log` its entire memory history.

Pulse
68%
Release
Apr 22, 2026
213
1
35%

PicoClaw

Go

A hyper-optimized Go rewrite of OpenClaw that runs on $10 hardware with under 10MB RAM—perfect for embedded enthusiasts and resource-constrained deployments. The project has exploded in popularity, hitting 12K stars in a single week with an active community pushing rapid feature development.

Pulse
88%
Release
Apr 22, 2026
28,441
15
35%

CoPaw

Python

Alibaba's ambitious open-source personal AI workstation that positions itself as a feature-rich OpenClaw alternative with multi-channel support (Discord, Telegram, Feishu, DingTalk, iMessage). Built on the AgentScope framework, it emphasizes local/cloud flexibility and persistent autonomous operation.

Pulse
78%
Release
Apr 22, 2026
15,802
19
35%

OpenClaw

TypeScript

The OG personal AI assistant that sparked the entire self-hosted agent movement—OpenClaw is the 300K-star juggernaut that makes ChatGPT look like a toy. It's chaotic, powerful, and slightly dangerous, with a community that's equal parts evangelists and security researchers sounding alarms.

Pulse
72%
Release
Apr 22, 2026
362,432
18
35%

nanobot

Python

An ultra-lightweight personal AI assistant that delivers core agent functionality with 99% fewer lines of code than OpenClaw. Perfect for developers who want to actually understand every line of their AI assistant's codebase.

Pulse
82%
Release
Apr 21, 2026
40,540
7
35%

LobsterAI

TypeScript

NetEase Youdao's polished take on the OpenClaw ecosystem, delivering an Electron-based desktop agent with robust Chinese platform integrations like DingTalk and Feishu. Its Cowork mode and Alpine Linux sandboxing make it a compelling option for users who want supervised task execution with enterprise-grade local data control.

Pulse
68%
Release
Apr 21, 2026
5,085
21
35%

AionUi

TypeScript

AionUi is a cross-platform 'Cowork' desktop app where AI agents work alongside users—reading files, writing code, and automating tasks with full transparency. It differentiates itself with zero-setup built-in agents, multi-agent orchestration, and 24/7 automation capabilities.

Pulse
68%
Release
Apr 21, 2026
22,388
20
35%

IronClaw

Rust

IronClaw is the security-hardened, privacy-first alternative to OpenClaw that runs everything in WASM sandboxes and keeps your data local. Built in Rust with defense-in-depth architecture, it's for users who want agentic AI power without the credential-leaking, prompt-injection nightmares of mainstream alternatives.

Pulse
87%
Release
Apr 21, 2026
11,917
21
35%
Early Signal

On Our Radar

Recently discovered clones under observation. These haven't been fully verified yet — we're watching for activity, community growth, and legitimacy before adding them to the main tracker.

34 Watching

NeuroClaw

ai-labs/neuroclaw

Promotion Candidate
892
🔤Rust
healthy

High-performance AI agent runtime built in Rust with zero-copy message passing and advanced sandboxing capabilities.

67 days
v0.2.1-alpha

Kai

SimonSchubert/Kai

Observing
569
🔤Kotlin
healthy

OpenClaw alternative in your pocket

48 days
v2.3.2

Hermitclaw

brendanhogan/hermitclaw

Observing
321
🔤Python
warning

hermitclaw - a new project under observation.

49 days

Gitclaw

SawyerHood/gitclaw

Observing
255
🔤Python
warning

OpenClaw but it runs entirely on github actions

34 days

Autoclaw

tsingliuwin/autoclaw

Observing
211
🔤TypeScript
warning

AutoClaw is a hyper-lightweight AI agent designed to live inside Docker containers. Unlike heavy, GUI-dependent agents, AutoClaw is built for headless, massive-scale concurrency.

41 days

Golem

MEKXH/golem

Observing
180
🔤Go
healthy

Golem — A self-hosted personal AI agent built in pure Go. Terminal UI +multi-channel bot, multi-LLM support, tool calling, long-term memory, skill packs, and cron scheduling. Single binary. 纯 Go 打造的可自托管个人 AI 智能体。终端 TUI + 多渠道机器人适配 ,多模型适配,工具调用,长期记忆,Skills扩展,内置定时任务。单文件部署,召之即来。

45 days
v0.7.1

Openmozi

oujingzhou/openmozi

Observing
179
🔤TypeScript
healthy

轻量级的clawdbot/openclaw,支持飞书,钉钉,QQ和企业微信

45 days
1.11.1

TestClaw

testorg/testclaw

Observing
127
🔤Python
healthy

A brand new experimental AI agent framework focusing on simplicity and rapid prototyping. Very early stage, but shows promise.

53 days
Page 1 of 5

Clones meeting our thresholds (★ 500+ · 60%+ Pulse · 14d+ observed) get promoted automatically

Situation Summary

Automated fallback summary

1 signals and 1 strong signals indicate above-average activity.

Day -6: elevated
Day -5: normal
Day -4: calm
Day -3: calm
Day -2: calm
Day -1: elevated
Day -0: elevated
7 Days Ago Today

Category Activity

SEC
3 Signals
INF
2 Signals
BIZ
1 Signals
Total Validated Signals 1
2026-W16 Briefing: Weekly briefing fallback
Open Dashboard

Security Radar

Security radar summary for OpenClaw, Hermes-agent, nanobot, ZeroClaw, AstrBot, PicoClaw.

  • OpenClaw: Sandboxing 4 of 10, API Security 5 of 10, Network Isolation 3 of 10, Telemetry Safety 7 of 10, Shell Protection 1 of 10.
  • Hermes-agent: Sandboxing 9 of 10, API Security 9 of 10, Network Isolation 6 of 10, Telemetry Safety 9 of 10, Shell Protection 8 of 10.
  • nanobot: Sandboxing 5 of 10, API Security 7 of 10, Network Isolation 6 of 10, Telemetry Safety 8 of 10, Shell Protection 4 of 10.
  • ZeroClaw: Sandboxing 8 of 10, API Security 7 of 10, Network Isolation 7 of 10, Telemetry Safety 9 of 10, Shell Protection 6 of 10.
  • AstrBot: Sandboxing 8 of 10, API Security 7 of 10, Network Isolation 6 of 10, Telemetry Safety 7 of 10, Shell Protection 7 of 10.
  • PicoClaw: Sandboxing 4 of 10, API Security 6 of 10, Network Isolation 4 of 10, Telemetry Safety 7 of 10, Shell Protection 3 of 10.

Evaluation Scale: 10 = Maximum Safety / 1 = High Risk