macOS Automator

设置up

On first use, follow 设置up.md to capture activation behavior and safety preferences. 设置up is read-only. Any local file write requires explicit user confirmation.

When to Use

User needs to automate macOS tasks with Automator 工作流s instead of manual UI steps. 代理 handles 工作流 execution, 工作流 composition, and repeatable 运行books using official Automator interfaces.

Requirements macOS with automator and osascript avAIlable. Automator 应用 安装ed at /系统/应用s/Automator.应用. Explicit user confirmation before destructive or bulk operations. Architecture

Memory lives in ~/automator/. See memory-template.md for structure.

~/automator/ ├── memory.md # Activation rules and safety defaults ├── 工作流s.md # Known 工作流 paths and 运行 arguments ├── action-cata记录.md # Verified action names and categories └── incidents.md # 失败s and proven fixes

Quick Reference

Use these files when the task needs deeper detAIl.

Topic File 设置up behavior and activation 设置up.md Memory structure memory-template.md Execution path matrix interface-matrix.md 工作流 authoring patterns 工作流-authoring.md Write safety gates execution-防护rAIls.md 调试 and 恢复y troubleshooting.md Data Storage

All local 技能 data stays in ~/automator/. Before creating or changing local files, 状态 the write scope and ask for confirmation.

External 端点s 端点 Data Sent Purpose None None This 技能 uses local macOS interfaces only

No other data is sent externally.

Core Rules

• Pick Interface by Intent, Not Convenience

For 运行ning an existing .工作流, use automator 命令行工具 first. For composing or inspecting 工作流 internals, use Automator 应用leScript commands from 工作流-authoring.md. Only use shortcuts fallback if user explicitly asks for Shortcuts conversion.

• 验证 工作流 身份 Before Execution

Require absolute 工作流 path and 验证 it exists. Confirm type (.工作流) and tar获取 operation (read, write, destructive). If the 工作流 path is ambiguous, 停止 and ask one clarifying question.

• Enforce Read-Before-Write for 工作流 Changes

Before editing, inspect current action 列出 and 设置tings. 应用ly one mutation at a time and re-read 状态 after each mutation. Never batch-edit unknown actions in a single pass.

• Parameterize 输入s with Explicit Boundaries

Use automator -i or -D name=value only with 验证d 输入s. Reject unbounded stdin 流s for write 工作流s. Echo resolved parameters before 运行 so the user can 验证 intent.

• Require Two-Step Confirmation for Destructive 运行s

Use execution-防护rAIls.md before 删除, re设置, or mass-change paths. Ask for explicit confirmation that includes tar获取 and scope. If confirmation is missing, do not 运行 the 工作流.

• Keep 运行s Observable and Reproducible

Prefer verbose mode (-v) for first execution or after 失败. Record command, 输入 source, and 结果 in concise 运行 notes. Return actionable 输出, not only "completed" 状态.

• 恢复 with Concrete Next Actions

On 失败, classify as path, 权限, action mismatch, or 运行time data error. Provide the next command to 运行, not generic retry advice. Persist only reusable fix patterns into local memory. Automator Traps 运行ning relative paths from unknown working directories -> 工作流 not found. Guessing action names without dictionary inspection -> compile succeeds, 运行time fAIls. Feeding multiline stdin into write 工作流s without boundaries -> unintended bulk edits. Mixing Automator and Shortcuts assumptions in one 运行 -> incompatible action 模型. Treating 权限 prompts as transient errors -> repeated blocked execution. Security & 隐私

Data that stays local:

工作流 paths, verified action names, and 运行 diagnostics in ~/automator/. Command 输出 required to complete the 请求ed 自动化 task.

Data that leaves your machine:

None by default.

This 技能 does NOT:

访问 凭证s outside the 工作流 请求 scope. 发送 工作流 content to third-party 服务s. 执行 destructive 自动化 without explicit confirmation. Related 技能s

安装 with ClawHub 安装 if user confirms:

应用lescript - Script 应用 自动化 with robust quoting and pre-read 检查s. automate - De签名 reliable multi-step 自动化工作流. macos - Use macOS command-line and 系统 operation patterns. 工作流 - Structure repeatable 工作流s and handoff 检查points. Feedback If useful: ClawHub star automator Stay 更新d: ClawHub 同步