ListenClaw

Formats responses for voice/audio output via the ListenClaw voice gateway. Use when: (1) A message starts with [ListenClaw] — this means the message was sent...

Security Scan

The skill's files and runtime instructions match its stated purpose (formatting text for a voice gateway); the installer simply pulls the SKILL.md from the project's GitHub and writes it into the agent workspace, and there are no credential or privileged actions requested.

Purpose & Capability

Name/description (formatting responses for ListenClaw) align with the SKILL.md instructions, which only constrain reply formatting for a voice gateway. No unrelated binaries, env vars, or capabilities are requested.

Instruction Scope

SKILL.md contains only formatting rules for spoken responses and activation rules (messages prefixed with [ListenClaw]). It does not instruct the agent to read unrelated files, access credentials, or send data to third-party endpoints.

Install Mechanism

install.sh uses curl to fetch SKILL.md from raw.githubusercontent.com (a standard release host) and writes it into ~/.openclaw workspace. No archives are extracted and no obscure host or URL shorteners are used.

Credentials

The skill requires no environment variables, credentials, or config paths. The installer writes into the agent's workspace, which is appropriate for a skill; there are no requests for unrelated secrets.

Persistence & Privilege

(always) is false and the skill is user-invocable. The installer writes its own SKILL.md into the agent workspace (normal). It does not modify other skills or request permanent elevated privileges.

Assessment

This skill appears coherent and low-risk: the installer only downloads the SKILL.md from the project's GitHub and places it in your agent workspace, and the runtime instructions only control how replies are spoken. Before installing, verify the GitHub repository is the expected one and inspect the fetched SKILL.md and any restart script (~/.openclaw/restart-gateway.sh) the installer asks you to run. Avoid running unknown installers that fetch and execute code from arbitrary URLs; in this case the URL is raw.githubusercontent.com (standard), and no credentials or system-wide changes are requested.

ListenClaw Voice Gateway

This skill activates when a message starts with [ListenClaw].

The message was sent by the ListenClaw voice gateway. The user spoke into a microphone — their speech was transcribed and forwarded here. Your reply will be converted back to speech by a TTS engine and played through the user's speakers or headphones.

Rules

• Reply in natural spoken language only
• No Markdown: no **bold**, no # headings, no - lists, no | tables |, no code blocks, no horizontal rules
• Express data as sentences: instead of a table, say the values out loud
• Keep sentences short so TTS can pause naturally between them
• No parenthetical notes, URLs, or technical symbols unless asked

Examples

Message: [ListenClaw] 北京今天天气怎么样

Wrong:

| 指标 | 数值 |
|------|------|
| 天气 | 晴   |
| 气温 | 25°C |

Correct:

北京今天晴天，气温25度，比较适合出门。

---

Message: [ListenClaw] set a reminder for tomorrow at 9am

Wrong:

**Done** — Reminder set:
- Time: 09:00 tomorrow
- Content: reminder

Correct:

Got it, I've set a reminder for tomorrow at 9am.