经 AI Skill Hub 精选评估,麦克利 获评「强烈推荐」。这款Agent工作流在功能完整性、社区活跃度和易用性方面表现出色,AI 评分 8.0 分,适合有一定技术背景的用户使用。
麦克利 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
麦克利 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 克隆仓库 git clone https://github.com/ljh-sh/macli cd macli # 查看安装说明 cat README.md # 按 README 完成环境依赖安装后即可使用
# 查看帮助 macli --help # 基本运行 macli [options] <input> # 详细使用说明请查阅文档 # https://github.com/ljh-sh/macli
# macli 配置说明 # 查看配置选项 macli --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export MACLI_CONFIG="/path/to/config.yml"
Minimal context with maximum flexibility — macOS system tools for AI agents. Native Apple frameworks. AI-friendly JSON/TSV output.
macli turns macOS system internals into a clean CLI. SMC sensors, streaming monitor, calendar/reminders, notifications — all callable from shell pipes or LLM agents, all JSON/TSV. One ~400 KB Swift binary. No Python runtime, no osascript overhead, no GUI.
Use it when you (or your AI agent) need to ask macOS something that system_profiler / ioreg / osascript either can't answer or answer badly: CPU die temperature right now, stream 1 Hz sensor readings into awk, today's calendar as JSON.
Mirrors: github.com/ljh-sh/macli · codeberg.org/ljh-sh/macli
Requires Swift 5.10+ / macOS 12+.
git clone https://github.com/ljh-sh/macli
cd macli
swift build -c release
Homebrew 6 added a trust step for third-party taps. Run brew trust ljh-sh/cli once, then brew install ljh-sh/cli/macli. This is a security feature, not a bug.
Mostly yes.
SOURCE_DATE_EPOCH, ZERO_AR_DATE, deterministic mtimes, RPATH removal — lives in .x-cmd/release.common.sh.macli ships with ad-hoc signature (no Apple Developer ID). For direct-download installs, strip the quarantine attribute:
xattr -dr com.apple.quarantine /usr/local/bin/macli The Homebrew formula does this automatically via post_install.
macOS TCC is prompting for Calendar/Reminders access. Click the system dialog to grant. Subsequent calls are instant. If you missed the prompt, go to System Settings → Privacy & Security → Calendars (or Reminders) and enable the terminal you're running macli from.
Expected. smc86 queries the Intel-Mac SMC key space, which Apple cleared on Apple Silicon. Use macli smc (not smc86) on M-series Macs.
osascript routes through AppleScript → Calendar.app RPC channel → permission prompts. Every cold start loads the AppleScript component. macli links EventKit.framework directly and requests permission once via the standard TCC prompt; subsequent calls are in-process.
Read commands (cal ls, event ls) never touch state. Write commands (cal add, event add, reminder add) only run when you invoke them explicitly, with the exact arguments you pass. macli never syncs, never deletes, never auto-modifies.
Short version: macOS won't let a bare CLI use speech recognition.
SFSpeechRecognizer needs an Info.plist with NSSpeechRecognitionUsageDescription..app, which isn't worth the pipeline complexity.Both supported via a single universal binary. On Apple Silicon use macli smc; on Intel use macli smc86. The binaries are identical; the subcommand selects the sensor path.
高质量的开源AI工作流工具
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
AI Skill Hub 点评:麦克利 的核心功能完整,质量优秀。对于自动化工程师和运维人员来说,这是一个值得纳入个人工具库的选择。建议先在非生产环境试用,再逐步推广。
| 原始名称 | macli |
| 原始描述 | 开源AI工作流:0.1MB macOS system tools for AI Agents. Native Apple frameworks. AI friendly wit。⭐8 · Swift |
| Topics | AImacOSSwift工作流 |
| GitHub | https://github.com/ljh-sh/macli |
| License | Apache-2.0 |
| 语言 | Swift |
收录时间:2026-06-16 · 更新时间:2026-06-16 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端