Walkie-Talkie Mode
PassAudited by ClawScan on May 1, 2026.
Overview
This is a coherent WhatsApp voice-reply workflow, with review-worthy notes because it sends replies through the messaging tool and relies on local helper executables that are not included in the package.
Use this only if you are comfortable with the agent transcribing WhatsApp voice messages, invoking the named local audio tools, and sending both text and voice replies. Verify the local tools before enabling the mode.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
When enabled, the agent can send voice-note replies, and the instructions also call for text replies, into the WhatsApp conversation.
This gives the skill a route to send media through the active messaging identity. That is expected for a WhatsApp walkie-talkie mode, but users should notice the delegated messaging authority.
Then send `/tmp/reply.ogg` via `message` tool with `filePath`.
Use the skill only with the intended account and chat, and disable it when automatic voice replies are not desired.
The actual transcription and TTS behavior depends on local tools that are not included here.
The runtime instructions reference a helper script and local binary while the supplied package contains only SKILL.md and no install spec. This is purpose-aligned, but those components are outside the reviewed artifact set.
Use `tools/transcribe_voice.sh` to get the text. ... generate speech using `bin/sherpa-onnx-tts`.
Before using the skill, confirm that the referenced local script and binaries are trusted and are the ones you expect.
The agent may run the named local TTS/transcription commands and write a temporary audio file for sending.
The skill includes local executable invocation and temporary audio-file output. This is central to the stated local TTS purpose and is disclosed.
bin/sherpa-onnx-tts /tmp/reply.ogg "Tu mensaje aquí"
Ensure the executable paths resolve to trusted local tools and that temporary audio output in /tmp is acceptable.