Five months. That’s all it took for OpenClaw to go from a quietly published GitHub repository to one of the fastest-growing open-source projects in history. As of late March 2026, the main openclaw/openclaw repository sits at approximately 337,000 GitHub stars and 66,000 forks — numbers that would be remarkable for a decade-old project, let alone one that didn’t exist until November 2025.
A Brief History of Names
OpenClaw was created by Austrian developer Peter Steinberger and originally published as Clawdbot in November 2025. It ran as a single Node.js gateway process on 127.0.0.1:18789, connecting your AI models of choice to over 50 messaging platforms: WhatsApp, Telegram, Slack, Discord, Signal, iMessage, WeChat, Matrix, and more.
On January 27, 2026, Anthropic raised trademark concerns about the name’s similarity to “Claude,” and the project was briefly renamed Moltbot. Three days later, it became OpenClaw — the name that stuck. By February 2026, the project had grown large enough that Steinberger transferred governance to the newly formed OpenClaw Foundation.
The speed of adoption was staggering. The project reportedly hit 100,000 stars in its first two days of viral attention — a rate that puts it in the company of projects like Docker and Node.js in terms of early traction.
Ecosystem Scale: The Numbers That Matter
The star count is just the headline figure. The real story is the breadth of ecosystem activity that has grown up around it:
- ~66,000 forks of the main repository
- 5,700+ skills on the ClawHub registry (clawhub.ai) — the official “App Store” for OpenClaw extensions
- 20+ npm packages in the official ecosystem, including official channel plugins from Tencent and Feishu
- 10+ community awesome-lists curating the best resources
- 52+ alternatives tracked on Shelldex.com alone
- Official support from enterprise players: NVIDIA, Tencent, Cloudflare, and ByteDance
The openclaw GitHub organization itself hosts 23 repositories, including the core gateway, the ClawHub skill registry, the acpx headless client, the lobster workflow shell, and a Windows companion app.
Category Breakdown
The forks and variants fall into seven recognizable categories:
Language Rewrites (Performance & Security)
The largest and most active category. Developers concerned with OpenClaw’s TypeScript/Node.js footprint (~390 MB RAM at runtime) have produced rewrites in Rust, Go, and Zig. ZeroClaw (Rust) achieves a 3.4 MB binary and under 5 MB RAM. NullClaw (Zig) goes further: a 678 KB binary with sub-8ms boot times. PicoClaw (Go, from hardware company Sipeed) targets $10 hardware like the Raspberry Pi Zero 2 W.
Enterprise & Managed Distributions
NVIDIA released NemoClaw in March 2026 — an early preview of an enterprise distribution with policy-based sandboxing and governance tooling. Cloudflare published moltworker as a proof of concept for running OpenClaw on Cloudflare Workers. Tencent ships QClaw as an Electron desktop app with WeChat OAuth. Managed hosting providers ClawTrust and Blink Claw offer SaaS deployments.
Security-Focused Forks
Following CVE-2026-25253 (CVSS 8.8) and the GhostClaw malware incident, a wave of security-hardened variants emerged: NanoClaw (container-isolated execution), Carapace (fail-closed authentication, signed plugins), IronClaw (WebAssembly sandboxing), and SecureClaw/ClawSec as security plugin suites.
Embedded & Mobile
Six or more Android ports cover the spectrum from Termux CLI setups to full native APKs. The Raspberry Pi ecosystem has its own dedicated installer supporting seven different OpenClaw flavors.
Regional & Localization
A rich Chinese-language ecosystem includes official Tencent WeChat and WeCom channel plugins, official Feishu/Lark support from ByteDance, and multiple community localization forks with hourly upstream sync.
Multi-Agent Coordination
ClawTeam (HKU Data Intelligence Lab), ClawSwarm, and ClawWork represent a sub-ecosystem building agent orchestration on top of OpenClaw — including one project where AI agents must earn income to remain economically solvent.
Tooling & Infrastructure
ClawRouter (BlockRunAI) routes requests to the cheapest capable model in under 1ms, claiming 92% API cost savings. The OpenClaw Foundation’s own lobster workflow shell and the ClawHub CLI round out the official tooling story.
Top 5 Implementations by Stars
╔══════════════════════════════════╦═══════════╦════════════╦══════════╗ ║ Project ║ Language ║ Stars ║ RAM ║ ╠══════════════════════════════════╬═══════════╬════════════╬══════════╣ ║ openclaw/openclaw ║ TypeScript║ ~337,000 ║ ~390 MB+ ║ ║ qwibitai/nanoclaw ║ TypeScript║ ~21,500 ║ ~800 KB ║ ║ sipeed/picoclaw ║ Go ║ ~12,000+ ║ <10 MB ║ ║ nullclaw/nullclaw ║ Zig ║ ~6,830 ║ ~1 MB ║ ║ BlockRunAI/ClawRouter ║ — ║ ~4,300 ║ — ║ ╚══════════════════════════════════╩═══════════╩════════════╩══════════╝
What Comes Next
The OpenClaw ecosystem is moving faster than almost any open-source project in recent memory. New variants appear weekly, enterprise entrants are just beginning to find their footing, and the security landscape is still maturing after a rough start in early 2026.
ClawTrackr exists to make sense of all of it. In the articles that follow, we dig into the language rewrite wars, the enterprise entrants, the security incidents, and the practical question every developer with a spare Raspberry Pi is asking: which variant should I actually run?