Overview

AgentZero is a lightweight, security-first agent runtime and CLI built entirely in Rust. It provides a single-binary runtime with streaming output, trait-driven extensibility, encrypted storage, and fail-closed security defaults.

Why AgentZero

Security that ships with the product. Scoped tools, allowlists, audit logging, and encrypted storage are built into the core — not bolted on as plugins. Every agent runs with fail-closed defaults from day one.
Zero vendor lock-in. Swap LLM providers, memory backends, tools, and communication channels through clean trait interfaces. Your agent logic stays the same whether you’re running GPT-4, Claude, or a local model.
One binary, deploy anywhere. A single native Rust binary gives you the full runtime — CLI, gateway, daemon, and tools. No containers, no interpreters, no external dependencies. Cold starts in milliseconds.
Built for production. OTP-gated tool execution, secret redaction, audit trails, and an emergency stop give you the operational controls that real deployments demand.

Features

Single Binary

One cargo install gives you the full runtime — CLI, gateway, daemon, and all tools. No Docker, no Node, no Python.

Trait-Driven Core

Provider, MemoryStore, Tool, and Channel are all traits. Swap implementations without touching orchestration code.

Secure by Default

Deny-by-default tool execution, filesystem scoping, shell command allowlists, secret redaction, and audit trails.

Fast Cold Starts

Native Rust binary starts in milliseconds. No JIT warmup, no interpreter overhead, no container boot.

Streaming Output

Stream agent responses token-by-token via CLI, SSE, or WebSocket. Real-time feedback for long-running tasks.

System Prompts

Configure a system prompt per agent or delegate to control personality, constraints, and behavior.

At a Glance


Language	100% Rust
Crates	16 workspace crates
CLI Commands	37+ subcommands
Providers	OpenAI-compatible (OpenRouter, OpenAI, Anthropic, Ollama, local)
Memory	SQLite (default), Turso/libsql (optional)
Tools	50+ built-in (file I/O, shell, networking, browser, delegation, memory, git, SOP, cron, hardware, MCP, and more)
Plugins	WASM sandbox with integrity verification
Security	Allowlists, OTP gates, audit trail, secret redaction, estop
License	MIT / Apache-2.0

Quick Start

# Build from source
cargo build -p agentzero --release

# Run interactive onboarding
cargo run -p agentzero -- onboard --interactive

# Set your API key
export OPENAI_API_KEY="sk-..."

# Send a message
cargo run -p agentzero -- agent -m "hello"

# Check system health
cargo run -p agentzero -- doctor models
cargo run -p agentzero -- status

See the full Installation guide and Quick Start for details.