← 返回 Skills 市场
stingan

Ceo Notify Agents

作者 Stingan · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
343
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install ceo-notify-agents
功能描述
Automation skill for Ceo Notify Agents.
使用说明 (SKILL.md)

name: ceo-notify-agents description: CEO向其他Agent发送通知(通过共享记忆文件) triggers:

  • "通知 {targetAgents} {message}"

  • "告诉 {targetAgents} {message}"

  • "提醒 {targetAgents} {message}"

  • "给 {targetAgents} 发消息 {message}"

  • "请通知 {targetAgents} {message}"

  • "通知一下 {targetAgents} {message}"

  • "让 {targetAgents} 知道 {message}" requires: [] actions:

  • name: write_notification tool: exec params: command: | #!/bin/bash TARGETS="{{targetAgents}}" MESSAGE="{{message}}" NOTIFICATION_DIR="/Users/anran/Documents/openclaw/shared_memory/notifications" TIMESTAMP=$(date +%Y%m%d_%H%M%S)

    mkdir -p "$NOTIFICATION_DIR"
    
    IFS=',' read -ra AGENTS \x3C\x3C\x3C "$TARGETS"
    for raw in "${AGENTS[@]}"; do
      agent=$(echo "$raw" | xargs)
      echo "$TIMESTAMP: $MESSAGE" >> "$NOTIFICATION_DIR/${agent}.log"
    done
    
    echo "$TIMESTAMP: 通知 $TARGETS - $MESSAGE" >> "$NOTIFICATION_DIR/all.log"
    
    /Users/anran/.npm-global/bin/openclaw memory index --agent main
    
    echo "通知已写入共享记忆,目标Agent下次对话时将看到。"
    
安全使用建议
This skill writes notification files to hard-coded user paths and runs an absolute openclaw CLI binary—things that should be configurable. Before installing: verify the skill's source/author; do not install if you don't trust them. If you want to use it, ask the author to (1) replace hard-coded paths with configurable env vars or relative paths, (2) avoid absolute references to a single user's home and to ~/.npm-global, and (3) document the required runtime binary and permissions. Run the skill in a sandbox or test environment first to confirm it only writes the files you expect. If you cannot get a trustworthy author or safer config, treat this as risky because it can modify files and trigger indexing of agent memory.
功能分析
Type: OpenClaw Skill Name: ceo-notify-agents Version: 1.0.0 The skill 'ceo-notify-agents' in SKILL.md contains a critical command injection vulnerability because it directly embeds unsanitized user input ({{targetAgents}} and {{message}}) into a bash script executed via the 'exec' tool. While the intended purpose of inter-agent communication via shared memory files appears functional, the implementation allows for arbitrary shell command execution. The script also relies on hardcoded absolute paths (e.g., /Users/anran/...) which is a security anti-pattern and indicates a brittle, environment-specific configuration.
能力评估
Purpose & Capability
The declared purpose is to notify other agents by writing shared-memory notification files, and the SKILL.md does exactly that. However the implementation uses hard-coded, user-specific absolute paths (/Users/anran/...) and a fixed binary location for openclaw, which is not coherent for a reusable or general-purpose skill.
Instruction Scope
Runtime instructions run a shell script that creates files under a specific Documents path and appends messages to per-agent logs and an all.log, then calls an absolute openclaw CLI path to index memory. These actions are within the stated notification scope but reference specific filesystem locations and execute a binary at a hard-coded path—both of which expand privilege and risk and are not declared or configurable in the skill.
Install Mechanism
This is an instruction-only skill with no install spec or downloaded code, which minimizes install-time risk. Nothing is written by an installer, but the runtime exec action will write files when invoked.
Credentials
The skill requests no environment variables or credentials, yet it assumes access to /Users/anran/... and to /Users/anran/.npm-global/bin/openclaw. That mismatch (no declared required env/config but reliance on user-specific paths) is disproportionate and brittle; it could fail or inadvertently manipulate files in that user's home if run in a different environment.
Persistence & Privilege
The skill is not always-enabled and does not request elevated platform privileges. Its main persistent effect is writing notification files and invoking the OpenClaw memory indexer, which is consistent with its function but should be scoped via configuration rather than fixed paths.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install ceo-notify-agents
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /ceo-notify-agents 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of ceo-notify-agents skill. - Allows the CEO to send notifications to other Agents via shared memory files. - Supports multiple trigger phrases in natural Chinese for flexible invocation. - Notifications are written into dedicated log files per target Agent and an overall log. - Ensures that notifications are indexed for future retrieval by Agents.
元数据
Slug ceo-notify-agents
版本 1.0.0
许可证
累计安装 1
当前安装数 1
历史版本数 1
常见问题

Ceo Notify Agents 是什么?

Automation skill for Ceo Notify Agents. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 343 次。

如何安装 Ceo Notify Agents?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install ceo-notify-agents」即可一键安装,无需额外配置。

Ceo Notify Agents 是免费的吗?

是的,Ceo Notify Agents 完全免费(开源免费),可自由下载、安装和使用。

Ceo Notify Agents 支持哪些平台?

Ceo Notify Agents 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Ceo Notify Agents?

由 Stingan(@stingan)开发并维护,当前版本 v1.0.0。

💬 留言讨论