← 返回 Skills 市场

Voice Wake Say TTS Responses (Native)

Name: Voice Wake Say TTS Responses (Native)
Author: xadenryan

作者 xadenryan · GitHub ↗ · v0.1.0

cross-platform ⚠ suspicious

2593

总下载

当前安装

版本数

在 OpenClaw 中安装

/install clawdbot-skill-voice-wake-say

功能描述

使用说明 (SKILL.md)

Voice Wake Say

Overview

Use macOS say to read the assistant's response out loud whenever the conversation came from Voice Wake/voice recognition. Do not use the tts tool (it calls cloud providers).

When to Use `say` (CHECK EVERY MESSAGE INDIVIDUALLY)

IF the user message STARTS WITH: User talked via voice recognition on m3

Step 1: Acknowledge with say first (so the user knows you heard them)
Step 2: Then perform the task
Step 3: Optionally speak again when done if it makes sense

IF the user message does NOT start with that exact phrase

THEN: Do NOT use say. Text-only response only.

Critical:

Check EACH message individually — context does NOT carry over
The trigger phrase must be at the VERY START of the message
For tasks that take time, acknowledge FIRST so the user knows you're working

Workflow

Detect Voice Wake context

Trigger ONLY when the latest user/system message STARTS WITH User talked via voice recognition on m3
If the message instructs "repeat prompt first", keep that behavior in the response.

Prepare spoken text

Use the final response text as the basis.
Strip markdown/code blocks; if the response is long or code-heavy, speak a short summary and mention that details are on screen.

Speak with say (local macOS TTS)

printf '%s' "$SPOKEN_TEXT" | say

Optional controls (use only if set):

printf '%s' "$SPOKEN_TEXT" | say -v "$SAY_VOICE"
printf '%s' "$SPOKEN_TEXT" | say -r "$SAY_RATE"

Failure handling

If say is unavailable or errors, still send the text response and note that TTS failed.

安全使用建议

This skill appears to do what it says (pipe text to macOS 'say') but it has small metadata/instruction mismatches you should fix or verify before installing: (1) confirm the agent will run on macOS (the skill assumes 'say' exists) — if the agent runs in the cloud or on Linux, the command will fail or do nothing; (2) ask the skill author to declare the required binary ('say') and an OS restriction in the metadata; (3) note that SKILL.md references optional env vars (SAY_VOICE, SAY_RATE) that are not declared — confirm whether you need to set them; (4) test in a safe environment to ensure audio is played where you expect (local speaker vs remote host) and that the trigger phrase behavior is acceptable; and (5) be aware that autonomous invocation is allowed by default — while the trigger is narrow, consider whether other skills or system messages could craft messages that accidentally trigger spoken output.

功能分析

Type: OpenClaw Skill Name: clawdbot-skill-voice-wake-say Version: 0.1.0 The skill is designed to use the local macOS `say` command for text-to-speech under specific conditions. It explicitly avoids cloud-based TTS, which is a good security practice. The instructions for the agent are highly restrictive regarding when and what to speak, using the agent's processed final response as input to `say`. The command execution `printf '%s' "$SPOKEN_TEXT" | say` is a safe way to pass text to the `say` utility, preventing shell injection. There is no evidence of data exfiltration, malicious execution, persistence, or obfuscation. The minor prompt injection vector related to 'repeat prompt first' only influences the spoken output, not arbitrary command execution or data exfiltration, and is within the scope of the skill's intended function.

能力评估

ℹ Purpose & Capability

The skill's stated purpose (speaking responses on macOS using the built-in 'say' command) aligns with the runtime instructions. However, the metadata does not declare an OS restriction or list 'say' as a required binary — a mismatch: the skill will only work on macOS with the 'say' binary present.

ℹ Instruction Scope

The SKILL.md confines actions to checking the latest message for a precise trigger phrase and running a local 'printf | say' command; it does not instruct reading unrelated files or contacting external endpoints. It does reference optional environment variables (SAY_VOICE, SAY_RATE) that are not declared in the skill metadata. The strict requirement to only check the latest message and ignore context is unusual but not inherently malicious.

✓ Install Mechanism

Instruction-only skill with no install steps and no code files — low installation risk (nothing is written to disk).

ℹ Credentials

The skill declares no required environment variables or credentials (appropriate for a local TTS helper). It nonetheless references optional env names (SAY_VOICE, SAY_RATE) in SKILL.md that are not declared; this is an inconsistency to be corrected but not by itself suspicious.

✓ Persistence & Privilege

always is false and there is no install-time persistence or modification of other skills or system-wide configuration. Model invocation is allowed (normal); the skill can be called autonomously, but its trigger is narrowly scoped to a user message prefix.

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 clawdbot-skill-voice-wake-say
安装完成后，直接呼叫该 Skill 的名称或使用 /clawdbot-skill-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

v0.1.0

voice-wake-say v0.1.0 - Initial release. - Adds ability to speak responses on macOS using the built-in `say` command when triggered by specific voice recognition input. - Only activates when user message starts with "User talked via voice recognition on m3". - Ensures spoken acknowledgment before performing tasks, for voice-initiated requests. - Falls back to text-only response and notifies if `say` fails or is unavailable.

元数据

Slug clawdbot-skill-voice-wake-say

版本 0.1.0

许可证 —

累计安装 12

当前安装数 12

历史版本数 1

常见问题