ZeroClaw

ZeroClaw is an open-source, lightweight personal AI agent runtime that runs as a single Rust binary on local hardware.

ZeroClaw is an agent runtime that functions as a single Rust binary on your own machine, allowing you to use your own keys or local models. It connects to over 70 LLM providers, including local options like Ollama, and supports more than 30 communication channels such as Telegram, Discord, and email. The system prioritizes security through supervised autonomy, OS-level sandboxing, and command allowlists. It operates on ARM and x86 architectures, from Raspberry Pis to cloud VMs. Users define agent behavior through swappable tools, memory, and configuration, ensuring that automation remains under local control.

ZeroClaw at a glance

Best for: Software engineer
Runs on: Linux, macOS, Docker
Pricing: Free
Released: February 2026
Category: AI agent tool
Website: zeroclaw.com

Key features

Supervised autonomy with approval gates for risky actions
Cryptographic tool receipts using HMAC-SHA256
OS-level sandboxing via Landlock, Bubblewrap, or Docker
Support for 70+ LLM providers and 30+ communication channels
Lightweight Rust-based binary with low memory usage
Hardware-level interaction including GPIO, STM32, and Arduino

Use it for

Run personal AI assistants on local hardware or homelabs
Automate shell tasks with supervised, auditable permissions
Execute scheduled tasks and webhooks in a sandboxed environment
Bridge local models to messaging platforms like Signal or Matrix

Alternatives to ZeroClaw

NullClaw — High-performance single-binary execution engine for autonomous AI agents.
thClaws — Runs locally in native Rust with zero cloud lock-in and multi-provider support.
NanoClaw — Runs in isolated Docker containers with credential protection via OneCLI.
Agent Zero — It provides a transparent, controllable environment for autonomous AI agents.
zerostack — Minimal memory footprint and high CPU efficiency compared to JS-based agents.
GoClaw — It offers Go-native performance with multi-tenant isolation and 5-layer security.
MetaClaw — Continuously improves your personal agent from live chats without requiring a local GPU cluster.
NemoClaw — Provides managed security, networking, and lifecycle operations for AI agents.
SelfClaw — It aggregates multiple AI models with automatic fallback and search.
MateClaw — It secures data and workflows within a self-hosted, audit-ready environment.
VulnClaw — It uses LLM agents to autonomously execute end-to-end security testing loops.
Atomic Agent — It runs locally on your machine, ensuring data privacy and no token costs.

All ZeroClaw alternatives