Frequently Asked Questions

OpenClaw is a self-hosted, open-source personal AI super-agent platform created by Peter Steinberger. It is written in TypeScript, released under the MIT license, and has accumulated over 339,349 GitHub stars since its launch. The project began life as a WhatsApp relay bot in November 2025 and went through several name changes — Clawd, Clawdis, Clawdbot, Moltbot — before settling on OpenClaw in late January 2026.

At its core, OpenClaw gives you a personal AI agent that can interact with your tools, files, and services on your behalf. It runs on your own hardware, keeping your data local by default. The project's rapid growth has spawned an entire ecosystem of clones, forks, and reimplementations, which is exactly what Shelldex exists to catalog.

You can explore the original project on its Shelldex profile page or view the full ranking on the leaderboard.

The best alternative depends on your priorities. By raw popularity, the top five projects by GitHub stars are OpenClaw (339k), Nanobot (37k), ZeroClaw (29k), PicoClaw (26k), and AstrBot (28k). Each takes a different approach: Nanobot is a research-focused Python implementation in 4,000 lines, ZeroClaw is a Rust-based runtime with sub-10ms startup, and PicoClaw targets embedded hardware like $10 RISC-V boards.

If security is your primary concern, look at NanoClaw (26k stars, OS-level container isolation), IronClaw (Rust-hardened privacy fortress), or Moltis (zero-unsafe Rust with full audit trails). For cloud-native deployments, Moltworker runs on Cloudflare Workers. Check the leaderboard for the full ranked list.

Shelldex currently tracks 54 projects across the OpenClaw ecosystem. That number includes the original OpenClaw itself, direct forks, clean-room reimplementations in different languages, and inspired projects that have taken the personal AI agent concept in new directions.

The ecosystem is still growing rapidly. New projects surface weekly, and the submission page is open for anyone to add a project they have found or built. Browse the full directory to see every tracked project at a glance.

The OpenClaw ecosystem refers to the constellation of projects that have emerged around the original OpenClaw personal AI agent. It includes direct forks, language ports, security-hardened variants, minimal reimplementations, cloud-hosted versions, and entirely new frameworks that cite OpenClaw as their primary inspiration.

Projects in the ecosystem span 9 programming languages — TypeScript, Python, Rust, Go, C, Bash, and Zig — and range from BashoBot (a pure Bash implementation) to production-grade platforms like OpenFang (137,000 lines of Rust). Some focus on embedded hardware, others on enterprise cloud, and still others on security research. The analysis page breaks down the ecosystem's language distribution, activity trends, and growth patterns.

Yes. OpenClaw is released under the MIT license, one of the most permissive open-source licenses available. You can use, modify, and redistribute it freely, including in commercial products. The full source code is hosted on GitHub.

Most projects in the ecosystem also use permissive licenses. MIT is the most common, followed by Apache 2.0. A few alternatives use stronger copyleft licenses — for instance, TinyClaw uses GPL-3.0 and AstrBot uses AGPL-3.0. You can filter by license on the leaderboard to find projects that match your licensing requirements.

The lightest alternatives depend on what you mean by "light." In terms of resource footprint, ZeroClaw ships as a single 3.4MB Rust binary that starts in under 10 milliseconds and uses less than 5MB of RAM. PicoClaw goes even further, running on $10 RISC-V boards with under 10MB of RAM and a one-second boot time.

In terms of code complexity, BashoBot is the absolute minimum: a personal AI assistant built entirely in Bash using only standard Unix utilities like curl, jq, and base64. No Node.js, no Python, no compiled runtimes. TinyClaw and MicroClaw also target the minimal end of the spectrum, with MicroClaw adding Rust's memory safety guarantees.

Absolutely. Several projects in the ecosystem are specifically designed for resource-constrained hardware. PicoClaw was built from the ground up for embedded boards — it runs on $10 RISC-V hardware with less than 10MB of RAM, so a Raspberry Pi is more than capable. ZeroClaw's 3.4MB binary and sub-5MB RAM usage also make it an excellent fit for any Pi model.

BashoBot is another strong candidate since it requires only Bash 3.2+ and standard Unix tools — no heavy runtimes to install. For Rust enthusiasts, MicroClaw and IronClaw both compile to native ARM binaries. The main bottleneck on a Pi will be the LLM inference layer, not the agent framework itself, so pairing a lightweight agent with a cloud-hosted model API keeps resource usage manageable.

Rust is the second most popular language in the ecosystem after TypeScript, with 9 projects currently tracked by Shelldex. They span a wide range of use cases:

• ZeroClaw — Zero overhead, zero compromise
• OpenFang — The Agent Operating System
• IronClaw — Rust-hardened privacy fortress
• Moltis — Zero unsafe, full audit trail
• Spacebot — Concurrent multi-process AI agent for communities and teams
• MicroClaw — Rust agent inspired by NanoClaw
• OpenCrabs — Autonomous, self-improving Rust agent
• ZeptoClaw — Ultra-lightweight Rust agent with container isolation
• Carapace — Hardened Rust shell for autonomous agents

Rust's compile-time memory safety, absence of garbage collection pauses, and single-binary distribution make it particularly attractive for agent frameworks where security and performance matter. Visit the leaderboard to sort all projects by language.

Excluding the original OpenClaw (339k stars), the most-starred alternative is Nanobot with 36,742 stars. Nanobot is an ultra-minimal Python implementation designed for research and experimentation — just 4,000 lines of code that prioritize readability over feature density.

Close behind are ZeroClaw and PicoClaw, each with around 29,044 and 26,497 stars respectively. Rounding out the top five alternatives are AstrBot (28k) and AionUi (20k). The leaderboard provides real-time rankings with seven-day star growth trends to help you track which projects are gaining momentum fastest.

The 54 projects tracked by Shelldex span 9 programming languages. TypeScript leads with 19 projects, reflecting the ecosystem's web-first origins (OpenClaw itself is TypeScript). Python is next with 11 projects, favored by the research and automation communities. Rust has 9 projects, driven by developers who prioritize memory safety and performance.

Go accounts for 7 projects (including PicoClaw), while C has 3, Bash has 1, and Zig has 1. The C project targets bare-metal performance, BashoBot proves that a personal AI agent can run in pure shell script, and the Zig entry explores systems-level agent development without the complexity of C++. For a visual breakdown of language distribution across the ecosystem, see the analysis page.

OpenClaw is open-source software under the MIT license, and its code is publicly auditable on GitHub. Because it is self-hosted, your data stays on your own machine by default rather than being routed through a third-party cloud. That said, like any software that interacts with LLM APIs and executes actions on your behalf, it requires trust in the code you run and the plugins you install.

The community has taken security seriously. Several dedicated security-focused alternatives have appeared specifically to address potential attack surfaces. NanoClaw uses OS-level container isolation, Moltis enforces zero-unsafe Rust with a full audit trail, and Carapace requires signed WASM plugins. If security is a priority, these hardened alternatives are worth evaluating alongside the original.

Several projects compete for this title, each with a different security philosophy. NanoClaw takes a minimalism-first approach: five files, one process, OS-level container isolation. By reducing the surface area to the absolute minimum, there is less to exploit. It has 25,807 stars and is one of the most battle-tested security alternatives.

On the Rust side, IronClaw is a privacy-focused implementation with security hardening baked into its architecture. Moltis takes it further with zero-unsafe Rust and a complete audit trail for every agent action. Carapace adds signed WASM plugins and strict local-first defaults, preventing any untrusted code from executing without explicit cryptographic approval.

For users who want a corporate-backed option, Secure-OpenClaw by Composio offers a security-hardened fork with 500+ app integrations and enterprise-grade controls. The right choice depends on whether you prioritize minimal attack surface, memory safety, audit compliance, or managed security infrastructure.

Permission models vary significantly across the ecosystem. The original OpenClaw uses a skill-based permission system where each plugin declares the capabilities it needs, and the user approves or denies access. This is the most common pattern, and many forks inherit it directly.

Security-focused alternatives go further. NanoClaw runs each agent in an OS-level container, enforcing isolation at the kernel level rather than relying on application-layer checks. Moltis logs every action to an immutable audit trail, so you can retroactively inspect exactly what your agent did and why. Carapace requires plugins to be signed WASM modules, ensuring that only cryptographically verified code can run.

For zero-trust environments, SafeClaw takes the most radical approach: it uses no LLM at all, relying entirely on deterministic rule-based logic. This eliminates the entire class of prompt injection and hallucination-based permission bypasses. The trade-off is reduced flexibility, but the security guarantee is absolute.

It depends on your priorities. OpenClaw is the full-featured reference implementation with 339k stars, an enormous plugin ecosystem, and the largest community. If you want maximum functionality, the widest range of integrations, and the most community support, OpenClaw is the default choice.

NanoClaw strips that down to five files and one process, trading features for security and simplicity. It uses OS-level container isolation and has accumulated over 25,807 stars from developers who specifically value its security-first, radically minimal approach. If you do not need OpenClaw's full plugin surface and prefer a smaller, more auditable codebase, NanoClaw is the stronger choice.

Both are written in TypeScript and released under the MIT license, so switching between them does not require learning a new language or navigating license restrictions. For a side-by-side breakdown, see the comparison tool.

OpenClaw and ZeroClaw are fundamentally different projects that happen to solve the same problem. OpenClaw is a mature TypeScript platform with a massive ecosystem of plugins and integrations. ZeroClaw is a Rust-native runtime built for raw performance: sub-10ms startup, a 3.4MB binary, and less than 5MB of RAM at runtime.

Choose OpenClaw if you need the broadest plugin support, a large community for troubleshooting, and you are comfortable running a Node.js stack. Choose ZeroClaw if you are deploying to resource-constrained environments, need deterministic performance characteristics, or prefer Rust's compile-time safety guarantees. ZeroClaw's 29,044+ stars show significant community validation of the performance-first approach.

Use the comparison tool to see how they stack up on stars, activity, releases, and community health.

The term "fork" gets used loosely in this ecosystem. True forks — projects that share OpenClaw's actual codebase and diverge from it — include Secure-OpenClaw by Composio and Moltworker by Cloudflare. These inherit OpenClaw's TypeScript foundation and modify it for specific use cases (security hardening and serverless deployment, respectively).

Most other "alternatives" are better described as reimplementations or inspired projects. ZeroClaw, IronClaw, and Moltis are written from scratch in Rust. Nanobot is a clean-room Python implementation for research. PicoClaw targets a completely different platform (embedded Go on RISC-V boards). These projects share OpenClaw's concept of a personal AI agent but not its code.

Shelldex tracks all of these under one umbrella because the ecosystem is defined by the idea, not the codebase. The directory labels each project's category so you can quickly distinguish forks from ports from original works.

Shelldex is a community-maintained directory that catalogs every project in the OpenClaw ecosystem. It tracks clones, forks, reimplementations, and inspired projects — comparing them by GitHub stars, programming language, activity, and community health. Think of it as a species registry for every shell in the sea.

The site was formerly known as Clawdex (at theclawdex.com) before rebranding in February 2026. It is built with Astro and Tailwind CSS, hosted on GitHub Pages, and the entire codebase is open source. Every project listing is a YAML file in the repository, and enrichment data (stars, forks, recent commits) is pulled from the GitHub API at build time. Read more on the about page.

Head to the submit page for full instructions. The short version: every project in Shelldex is a YAML file in the GitHub repository. To add a project, open a pull request that adds a new YAML file to src/data/projects/ with the project's name, slug, description, language, GitHub URL, and status.

A human reviews every submission before it merges. The project must be an OpenClaw-inspired agentic framework — not a workflow, tutorial, or skill/plugin. Once merged, the site rebuilds automatically and the project appears in the directory, leaderboard, and comparison tools. No accounts, no forms, no approval committees. Just git.

Project data is enriched from the GitHub API on every production build. This means star counts, fork counts, open issues, latest releases, and commit timestamps are refreshed each time the site is deployed. Deployments happen regularly as new projects are submitted and reviewed.

New project additions depend on community submissions and maintainer curation. The ecosystem is growing quickly — Shelldex went from zero to 54 tracked projects within its first weeks. If you spot a project that should be listed, the submission process is open and straightforward.

No. Shelldex is an independent, community-maintained project created by Dave Onkels. It is not affiliated with OpenClaw, Peter Steinberger, or Anthropic. The directory exists to serve the community by mapping the ecosystem, not to endorse or rank any particular project.

Shelldex lists OpenClaw alongside every other project in the ecosystem under the same criteria: GitHub data, language, license, and activity. The about page has more details on the project's origins, naming history, and editorial approach.