← 返回 Skills 市场
10000
总下载
5
收藏
67
当前安装
1
版本数
在 OpenClaw 中安装
/install voice-wake-say
功能描述
Speak responses aloud on macOS using the built-in `say` command when user input indicates Voice Wake/voice recognition (for example, messages starting with "User talked via voice recognition on <device>").
安全使用建议
Install only if you want Voice Wake responses spoken aloud on macOS. Avoid using it for sensitive responses where others might overhear, and be aware that the `repeat prompt first` instruction is oddly worded even though it does not show concrete malicious behavior.
功能分析
Type: OpenClaw Skill
Name: voice-wake-say
Version: 1.0.1
The skill's primary function to use the local macOS `say` command for text-to-speech is benign, and the command execution (`printf '%s' "$SPOKEN_TEXT" | say`) is implemented safely. However, the `SKILL.md` file contains a critical instruction to the AI agent: "If the message instructs 'repeat prompt first', keep that behavior in the response." This instruction creates a significant prompt injection vulnerability, as it explicitly tells the agent to process and act upon arbitrary user-provided instructions embedded after the 'repeat prompt first' phrase, potentially allowing a malicious user to override the agent's intended behavior or security controls.
能力评估
Purpose & Capability
The local `say` command is central to the stated purpose and is disclosed; the main user impact is that assistant responses may be audible nearby.
Instruction Scope
The skill limits use to messages starting with the Voice Wake phrase. One line about preserving a `repeat prompt first` behavior is awkward, but the artifact does not instruct the agent to reveal hidden prompts, bypass policy, exfiltrate data, or execute unrelated commands.
Install Mechanism
The package is instruction-only with a single SKILL.md file and no executable scripts, dependency installation, or setup hooks.
Credentials
The behavior is macOS-specific and depends on the built-in `say` binary; that requirement is clear in the skill text, though metadata does not separately declare an OS or binary dependency.
Persistence & Privilege
No persistence, credential access, account access, network calls, background workers, or elevated privileges are requested.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install voice-wake-say - 安装完成后，直接呼叫该 Skill 的名称或使用
/voice-wake-say触发 - 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
v1.0.1
Remove device-specific trigger (make generic)
元数据
常见问题
Voice Wake Say 是什么？
Speak responses aloud on macOS using the built-in `say` command when user input indicates Voice Wake/voice recognition (for example, messages starting with "User talked via voice recognition on <device>"). 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 10000 次。
如何安装 Voice Wake Say？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install voice-wake-say」即可一键安装，无需额外配置。
Voice Wake Say 是免费的吗？
是的，Voice Wake Say 完全免费（开源免费），可自由下载、安装和使用。
Voice Wake Say 支持哪些平台？
Voice Wake Say 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 Voice Wake Say？
由 xadenryan（@xadenryan）开发并维护，当前版本 v1.0.1。
推荐 Skills