# 使用 Homebrew 安装 nullclaw 查看 nullclaw 的安装路径、可执行文件、元数据以及面向 AI 代理工作流的安全说明。 ## 安装 ```sh sudo av install brew:nullclaw ``` 其他安装命令: ### macOS - Homebrew (100%): ```sh brew install nullclaw ``` 证据: local Homebrew formula metadata ## 软件包事实 - **软件包键:** brew:nullclaw - **软件包管理器:** Homebrew - **版本:** 2026.5.29 - **来源摘要:** Tiny autonomous AI assistant infrastructure written in Zig - **主页:** - **仓库:** - **最后更新:** 2026-06-14T14:13:23-04:00 - **已生成:** 2026-08-03T19:37:03+00:00 ## 可执行文件 - nullclaw (别名) ## 安装行为 - Bottle: 不可用 ## 版本和新鲜度 - 页面生成时间: 2026-08-03 - 管理器版本: 2026.5.29 ## 项目历史与用法 NullClaw is a Zig-based autonomous AI assistant runtime packaged as a small single-binary tool. Its public materials emphasize low memory use, fast startup, provider agnosticism, and running agent infrastructure on inexpensive or resource-constrained machines. ### 项目历史 GitHub's repository API reports that nullclaw/nullclaw was created on February 16, 2026. That makes it a very young project in package-history terms, so there is not enough independent history to treat it like a long-established ecosystem component. The repository README frames the project as '100% Zig' assistant infrastructure with a 678 KB binary, roughly 1 MB RAM use, and startup under 2 ms. Those are project claims from the official repository, not independently benchmarked adoption facts. ### 采用历史 NullClaw's visible adoption is early but unusually loud for a 2026 repository. The GitHub API reported 7,736 stars and 907 forks on July 1, 2026, while the Homebrew formula showed stable version 2026.5.29 and 1,661 installs over its visible 365-day analytics window. The official docs focus on a source-build path with Zig, Git, and an LLM provider API key, followed by `nullclaw onboard`, `nullclaw agent`, `status`, `doctor`, and `capabilities --json`. That indicates adoption is aimed at agent-runtime experimenters and self-hosting users rather than a mature enterprise distribution channel. ### 使用方式 Practical use begins by building or installing the `nullclaw` binary, onboarding a provider such as OpenRouter, OpenAI, or Anthropic, then running an agent message or gateway. The docs also describe web-channel pairing through `~/.nullclaw/config.json`, which turns it from a terminal-only agent into a runtime that can talk to a browser UI. Package nerds would care about NullClaw as a counterpoint to Python/Node-heavy agent frameworks: it is distributed as a compiled Zig binary, it has minimal runtime assumptions, and its Homebrew formula depends on Zig only when building from source. ### 为什么软件包爱好者会关心 NullClaw is significant mostly as an early signal in AI-agent packaging: small native binaries and local runtimes are being used to package agent infrastructure. The history is too short for high confidence about durability. ### 时间线 - 2026-02-16: GitHub repository created. - 2026-05-29: Homebrew formula stable version visible as 2026.5.29. - 2026-07-01: GitHub API reported 7,736 stars and 907 forks; Homebrew formula reported 1,661 installs over 365 days. ### Related projects - NullHub - Zig - OpenRouter - OpenAI - Anthropic ### 来源 - - - - - ## 安全说明 formula declares a Homebrew service. infrastructure mutation or orchestration signal. - **Geiger 风险:** orange / 中 - formula declares a Homebrew service - infrastructure mutation or orchestration signal ## Configuration and credential file locations These source-backed paths show where this package keeps local settings or durable credentials. Automic Vault can use them as review targets for secret scanning, migration, and command approval. ## Configuration files - Unix: ~/.nullclaw/config.json ## Credential files - Unix: ~/.nullclaw/config.json ## Combined YAML source View the package source record on GitHub. [combined/nullclaw.yml](https://github.com/mxcl/pkgdb/blob/main/combined/nullclaw.yml) ## 来源 - pkg.so package database - Geiger risk classifier - curated configuration and credential file locations - curated package history - pkgdb category and tag curation - cross-ecosystem install command graph