🦀 ZeroClaw：31k Stars 的 Rust 自主 AI 助手运行时，一个二进制文件跑通全平台
项目地址：zeroclaw-labs/zeroclaw | ⭐ 31,224 | 🛠 Rust | 作者：ZeroClaw Labs
老实说，现在市面上 AI Agent 框架多得离谱。Python 的有 CrewAI、LangChain，Node 的有 ElizaOS，每个都要装一堆依赖、配一堆环境变量。踩过的坑都是泪——版本冲突、Python 3.11 不兼容某个包、pip install 跑十分钟崩了。最骚的是，你只是想有一个自己能控制的 AI 助手，不是搭一个微服务架构。
ZeroClaw 换了个思路：一个 Rust 二进制文件，下载就能跑，配置全在 ~/.zeroclaw/config.toml，没有运行时依赖。
一、安装：一条命令搞定
curl -fsSL https://raw.githubusercontent.com/zeroclaw-labs/zeroclaw/master/install.sh | bash
想自己编译也行：
git clone https://github.com/zeroclaw-labs/zeroclaw.git
cd zeroclaw
./install.sh
装完后 zeroclaw onboard 自动启动配置向导。
几个实用 flag：
./install.sh --prebuilt # 跳过编译提示，直接下二进制
./install.sh --source # 强制源码编译
./install.sh --minimal # 仅内核 (~6.6 MB)
./install.sh --skip-onboard # 只安装，稍后手动配置
二、启动：三步走
zeroclaw onboard # 向导：选模型供应商 + 配通信渠道
zeroclaw agent # 终端里直接聊天
zeroclaw service install # 注册为系统服务（systemd/launchctl/Windows Service）
配好之后，你的 AI 助手就跑起来了——支持 30+ 通信渠道（Discord、Telegram、Matrix、邮件、Webhook、CLI），20+ 模型供应商（Anthropic、OpenAI、Ollama、本地模型）。
三、配置：一个 TOML 文件搞定
# ~/.zeroclaw/config.toml
default_provider = "openai-codex"
default_model = "gpt-5-codex"
想加本地 Ollama 模型？
[providers.models.ollama]
api_url = "http://localhost:11434/v1"
api_key = ""
default_model = "llama-4"
每个通信渠道对应一个 [channels. 块。想接 Telegram？加个 [channels.telegram] 配 token 就行。
四、为什么选 ZeroClaw？
最骚的是，整个运行时就是一个 Rust 二进制，没有 npm node_modules、没有 Python virtualenv、没有 Docker。配好了就是你的。
总结
🦀 ZeroClaw: 31k Stars Rust-Powered Autonomous AI Assistant Runtime — One Binary, Every Platform
Project: zeroclaw-labs/zeroclaw | ⭐ 31,224 | 🛠 Rust | By: ZeroClaw Labs
Let's be honest — the AI agent ecosystem is a mess. Python frameworks need a virtualenv and 200MB of deps. Node ones dump 50,000 files into node_modules. Every framework wants you to build a microservice architecture just to chat with an LLM.
ZeroClaw takes a different approach: a single Rust binary with zero runtime dependencies. Download it, run it, configure it with one TOML file.
Install: One command
curl -fsSL https://raw.githubusercontent.com/zeroclaw-labs/zeroclaw/master/install.sh | bash
Or build from source:
git clone https://github.com/zeroclaw-labs/zeroclaw.git
cd zeroclaw
./install.sh
Useful flags:
./install.sh --prebuilt # skip build prompt, download binary directly
./install.sh --source # force source build
./install.sh --minimal # kernel only (~6.6 MB)
Quick Start: Three commands
zeroclaw onboard # setup wizard: choose provider + channels
zeroclaw agent # interactive chat in terminal
zeroclaw service install # register as system service
Config: One TOML file
# ~/.zeroclaw/config.toml
default_provider = "openai-codex"
default_model = "gpt-5-codex"
Each channel gets its own [channels. block. Provider configs go in [providers.models..
Why ZeroClaw stands out
The whole runtime is a single Rust binary. No npm, no venv, no Docker. Just one file that does everything.