iMessage Voice Reply
PassAudited by ClawScan on May 1, 2026.
Overview
The skill coherently generates local TTS audio for iMessage replies, but users should note that it can send messages through BlueBubbles and installs unpinned Python/model dependencies.
Install only if you want the agent to generate local TTS audio and send it through your BlueBubbles/iMessage setup. Review the recipient and content before sending, be aware that setup downloads unpinned packages and models, and consider disclosing that voice replies are synthesized.
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.
The agent can send an audio message from the user's messaging setup to a specified recipient.
The skill sends an attachment through the configured BlueBubbles/iMessage channel, which is expected for the purpose but uses delegated messaging authority from the user's account.
"action": "sendAttachment", "channel": "bluebubbles", "target": "+1XXXXXXXXXX", ... "asVoice": true
Use only with a trusted BlueBubbles configuration and confirm the intended recipient and content before sending.
Installation depends on current upstream packages and model downloads, which could change over time.
The setup script installs unpinned Python packages and then initializes/downloads Kokoro model assets. This is purpose-aligned for local TTS, but package and model provenance are not pinned by version or hash.
"$VENV_DIR/bin/pip" install -q kokoro-onnx soundfile numpy
If you need stronger reproducibility, pin dependency versions and verify model hashes before installing.
A recipient may interpret the generated audio as a normal recorded voice message.
The skill intentionally formats synthesized audio to look like a native iMessage voice bubble. This is disclosed and central to the feature, but recipients may not realize the audio was TTS-generated.
Voice messages appear as inline playable bubbles with waveforms — identical to voice messages recorded in Messages.app.
Consider disclosing when a voice reply is AI-generated, especially in sensitive conversations.