Token导航 LogoToken导航

Agent Skill · 研究检索

agent-notify(Agent 通知)

agent-notify 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

agent-notify 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为只读;许可证为 MIT-0,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 15,931。

总安装 7,508周安装 316GitHub Stars 公开资料未说明下载 2,629核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
0.1.0

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:agent-notify(Agent 通知)
来源仓库:https://github.com/miluer-tcq/agent-notify
安装命令:
openclaw skills install agent-notify
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install agent-notify

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

agent-notify 为 AI 编码代理提供跨平台通知声音与任务栏闪烁功能。适用于 Claude Code、OpenClaw、Cursor 等多种代理环境的通知提醒。支持播放警报声音与视觉效果,增强任务完成感知。

安装命令为 openclaw skills install agent-notify,需确认权限与维护状态。使用前请核实是否会触发联网、命令执行或文件读写操作。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install agent-notify;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

agent-notify 怎么安装?

在终端执行 openclaw skills install agent-notify。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

agent-notify 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

agent-notify 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

agent-notify 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

agent-notify 用的人多吗?

公开数据是累计安装 7,508 次,近一周安装 316 次,下载 2,629 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

agent-notify 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
agent-notify

Agent 通知 · 本页

7,508—只读MIT-0有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
Agent Notify
description
Cross-platform notification sound and taskbar flash for AI coding agents (Claude Code, OpenClaw, Codex, Kiro, Cursor, etc.). Plays alert sounds and visual notifications when the agent needs confirmation or completes tasks. Auto-detects the agent environment and configures hooks accordingly. Use this skill whenever the user mentions notification sounds, alert beeps, taskbar flash, completion alerts, or wants to know when the agent finishes or needs attention — even if they just say something vague like "I want a sound" or "how do I get notified". Also trigger on any mention of configuring, enabling, disabling, or customizing audio/visual alerts for any AI coding agent. Works on Windows, macOS, and Linux.
author
Miluer-tcq
version
1.0.0
license
GPL-3.0
triggers

Agent Notify Skill

You are an assistant that helps users configure cross-platform notification sounds and taskbar/dock alerts for AI coding agents. When triggered, you MUST follow the instructions below precisely.

Language Detection

Detect the user's language from their message. If they write in Chinese, respond in Chinese. If they write in English, respond in English. Match the user's language throughout the entire interaction.

Step 1: Offer Two Modes

Present two options to the user:

English:

🔔 Agent Notify — Setup

Choose a setup mode:

1️⃣  Quick Setup (recommended) — One-click install with default settings
    • Notification on: confirmation requests + task completion
    • Default system sounds
    • Taskbar flash: 5 times

2️⃣  Custom Setup — Choose your own settings
    • Pick which events trigger notifications
    • Configure custom sound files
    • Adjust taskbar flash count

Enter 1 or 2:

Chinese:

🔔 Agent Notify — 配置向导

请选择配置模式:

1️⃣  一键配置(推荐)— 使用默认设置快速安装
    • 通知触发:确认请求 + 任务完成
    • 默认系统提示音
    • 任务栏闪烁:5 次

2️⃣  自定义配置 — 自由选择设置
    • 选择哪些事件触发通知
    • 配置自定义提示音文件
    • 调整任务栏闪烁次数

请输入 1 或 2:

Step 2A: Quick Setup

If user chooses 1, proceed directly to Step 3: Install.

Use these defaults:

  • Hooks: Notification (type: confirm), Stop (type: done)
  • Sounds: system defaults
  • Taskbar flash count: 5

Step 2B: Custom Setup

If user chooses 2, walk through these options interactively:

2B-1: Select Hook Events

Present available hook events:

Available notification triggers:

[1] ✅ Confirmation requests (Notification hook) — when Claude asks for permission
[2] ✅ Task completion (Stop hook) — when Claude finishes responding
[3] ❌ Error alerts — when a tool call fails
[4] ❌ Pre-tool notification (PreToolUse) — before each tool execution
[5] ❌ Post-tool notification (PostToolUse) — after each tool execution

Enter numbers to toggle (e.g., 1,2,3), or press Enter to keep defaults:

2B-2: Custom Sounds

Ask if user wants custom sound files:

Custom sound files (leave blank for system defaults):

• Confirmation sound path: [blank = system default]
• Completion sound path: [blank = system default]
• Error sound path: [blank = system default]

Supported formats:
  Windows: .wav
  macOS: .aiff, .mp3, .wav
  Linux: .oga, .wav, .mp3

2B-3: Taskbar Flash Count

Taskbar flash count [default: 5]:

Step 3: Install

After collecting settings (default or custom), perform the installation:

3.1 Detect OS and Agent Environment

Run this command to detect the operating system:

uname -s 2>/dev/null || echo "Windows"
  • If output contains Darwin → macOS
  • If output contains Linux → Linux
  • If output contains MINGW, MSYS, CYGWIN, or Windows → Windows

Then detect which AI coding agent the user is running. This determines the config directory and hooks format:

# Detect agent config directory (first existing one wins)
for dir in ~/.claude ~/.codex ~/.openclaw ~/.kiro ~/.cursor; do
  if [ -d "$dir" ]; then echo "$dir"; break; fi
done

On Windows, also check:

for dir in "$USERPROFILE/.claude" "$USERPROFILE/.codex" "$USERPROFILE/.openclaw"; do
  if [ -d "$dir" ]; then echo "$dir"; break; fi
done

Set AGENT_HOME to the detected directory (e.g., ~/.claude). If none found, default to ~/.claude and create it.

The supported agents and their known config paths:

AgentConfig DirSettings FileHooks Support
Claude Code~/.claudesettings.json✅ Full (Notification, Stop, PreToolUse, PostToolUse)
OpenClaw~/.openclawsettings.json✅ Same hooks format as Claude Code
Codex (OpenAI)~/.codexsettings.json⚠️ May differ — check docs
Kiro (AWS)~/.kirosettings.json⚠️ May differ — check docs
Cursor~/.cursorvaries⚠️ Different hooks system

If the detected agent is not Claude Code or OpenClaw, warn the user that hooks compatibility is not guaranteed and ask if they want to proceed. The notification scripts themselves (sound + taskbar flash) work universally — only the hooks integration is agent-specific.

3.2 Copy Notification Script

Find the skill directory by searching for this skill.md file:

for base in ~/.claude/skills ~/.codex/skills ~/.openclaw/skills ~/.kiro/skills ~/.cursor/skills; do
  found=$(find "$base" -name "skill.md" -path "*/agent-notify/*" 2>/dev/null | head -1)
  [ -n "$found" ] && echo "$(dirname "$found")" && break
done

On Windows (Git Bash / MSYS2), also try:

for base in "$USERPROFILE/.claude/skills" "$USERPROFILE/.codex/skills" "$USERPROFILE/.openclaw/skills"; do
  found=$(find "$base" -name "skill.md" -path "*/agent-notify/*" 2>/dev/null | head -1)
  [ -n "$found" ] && echo "$(dirname "$found")" && break
done

If the find command fails or returns nothing, ask the user where they installed the skill.

Copy the script to $AGENT_HOME/ (not hardcoded ~/.claude/):

For Windows:

cp "<skill_dir>/scripts/notify-windows.ps1" "$AGENT_HOME/notify.ps1"

For macOS:

cp "<skill_dir>/scripts/notify-macos.sh" "$AGENT_HOME/notify.sh"
chmod +x "$AGENT_HOME/notify.sh"

For Linux:

cp "<skill_dir>/scripts/notify-linux.sh" "$AGENT_HOME/notify.sh"
chmod +x "$AGENT_HOME/notify.sh"

3.3 Write Config

Write the user's configuration to $AGENT_HOME/notify-config.json:

Construct the config JSON from the user's chosen settings and write it. For example, with defaults:

cat > "$AGENT_HOME/notify-config.json" << 'EOF'
{
  "hooks": {
    "Notification": { "enabled": true, "type": "confirm" },
    "Stop": { "enabled": true, "type": "done" }
  },
  "sounds": {
    "confirm": "",
    "done": "",
    "error": ""
  },
  "taskbar": {
    "flashCount": 5,
    "enabled": true
  }
}
EOF

Substitute actual values from user's choices (sound paths, flash count, enabled hooks).

3.4 Update Agent Settings

Read the existing $AGENT_HOME/settings.json file, then merge the hooks configuration into it.

IMPORTANT: Do NOT overwrite existing settings. Merge the hooks key into the existing JSON, preserving all other fields (env, permissions, model, statusLine, etc.).

Build hook commands based on OS. Replace $AGENT_HOME with the actual detected path in the commands below:

Windows hooks:

{
  "hooks": {
    "Notification": [
      {
        "matcher": "",
        "hooks": [
          {
            "type": "command",
            "command": "powershell.exe -ExecutionPolicy Bypass -File \"$AGENT_HOME/notify.ps1\" -Type confirm -ConfigPath \"$AGENT_HOME/notify-config.json\""
          }
        ]
      }
    ],
    "Stop": [
      {
        "matcher": "",
        "hooks": [
          {
            "type": "command",
            "command": "powershell.exe -ExecutionPolicy Bypass -File \"$AGENT_HOME/notify.ps1\" -Type done -ConfigPath \"$AGENT_HOME/notify-config.json\""
          }
        ]
      }
    ]
  }
}

macOS/Linux hooks:

{
  "hooks": {
    "Notification": [
      {
        "matcher": "",
        "hooks": [
          {
            "type": "command",
            "command": "bash $AGENT_HOME/notify.sh confirm $AGENT_HOME/notify-config.json"
          }
        ]
      }
    ],
    "Stop": [
      {
        "matcher": "",
        "hooks": [
          {
            "type": "command",
            "command": "bash $AGENT_HOME/notify.sh done $AGENT_HOME/notify-config.json"
          }
        ]
      }
    ]
  }
}

If user selected additional hooks in custom mode (error, PreToolUse, PostToolUse), add those hook entries as well with the appropriate type parameter.

For agents that don't support the Claude Code hooks format, inform the user and offer to just install the notification scripts manually. They can then integrate with their agent's own hook/event system.

3.5 Test

Run a test notification:

Windows:

powershell.exe -ExecutionPolicy Bypass -File "$AGENT_HOME/notify.ps1" -Type confirm -ConfigPath "$AGENT_HOME/notify-config.json"

macOS/Linux:

bash "$AGENT_HOME/notify.sh" confirm "$AGENT_HOME/notify-config.json"

3.6 Report Success

English:

✅ Agent Notify installed successfully!

Agent: [detected agent name]
Config: $AGENT_HOME/

Configured notifications:
  • 🔔 Confirmation requests → [sound type]
  • ✅ Task completion → [sound type]

Files installed:
  • $AGENT_HOME/notify.[ps1|sh]
  • $AGENT_HOME/notify-config.json
  • $AGENT_HOME/settings.json (hooks added)

⚠️  Please restart your agent for hooks to take effect.

To uninstall, say: "disable notification" or "关闭提示音"
To reconfigure, say: "configure notify" or "配置提示音"

Chinese:

✅ Agent Notify 安装成功!

运行环境:[检测到的 agent 名称]
配置目录:$AGENT_HOME/

已配置的通知:
  • 🔔 确认请求 → [提示音类型]
  • ✅ 任务完成 → [提示音类型]

已安装文件:
  • $AGENT_HOME/notify.[ps1|sh]
  • $AGENT_HOME/notify-config.json
  • $AGENT_HOME/settings.json(已添加 hooks)

⚠️  请重启你的 agent 使 hooks 生效。

卸载请说:"关闭提示音" 或 "disable notification"
重新配置请说:"配置提示音" 或 "configure notify"

Important Notes

  • $AGENT_HOME refers to the detected agent config directory (e.g., ~/.claude, ~/.codex, ~/.openclaw). Always use the actual resolved path in commands, not the variable name.
  • On Windows, $HOME in hook commands is expanded by the shell (Git Bash/MSYS2). If the user's environment doesn't expand $HOME, use the full absolute path instead (e.g., C:/Users/username/.claude/notify.ps1).
  • On macOS, the Dock bounce uses Python's AppKit.NSApplication.requestUserAttention_() which is available by default. If PyObjC is not available, the fallback printf '\a' (terminal bell) will trigger a Dock bounce in most terminal apps.
  • On Linux, notify-send handles both the desktop notification and the visual alert. If it's not installed, only the sound will play.
  • For agents other than Claude Code / OpenClaw, the hooks format may differ. If auto-configuration fails, the notification scripts can still be used standalone — just call them directly from whatever hook/event system the agent provides.

Uninstall / Disable

When user triggers with disable-related keywords (关闭提示音, disable notification, etc.):

  1. Detect $AGENT_HOME using the same logic as Step 3.1
  2. Read $AGENT_HOME/settings.json
  3. Remove ALL hook entries that reference notify.ps1 or notify.sh
  4. Write back the cleaned settings.json
  5. Remove $AGENT_HOME/notify.ps1 or $AGENT_HOME/notify.sh
  6. Remove $AGENT_HOME/notify-config.json
  7. Confirm:

English:

✅ Agent Notify has been uninstalled.

Removed:
  • $AGENT_HOME/notify.[ps1|sh]
  • $AGENT_HOME/notify-config.json
  • Notification hooks from $AGENT_HOME/settings.json

⚠️  Please restart your agent for changes to take effect.

Chinese:

✅ Agent Notify 已卸载。

已移除:
  • $AGENT_HOME/notify.[ps1|sh]
  • $AGENT_HOME/notify-config.json
  • $AGENT_HOME/settings.json 中的通知 hooks

⚠️  请重启你的 agent 使更改生效。

平台分布

OpenClaw

93.73%
按下载量换算2,464

安全审计

VirusTotal

通过

ClawScan

可疑

Static analysis

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills