Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Connects OpenClaw agents to Signal messenger via signal-cli. Handles sending and receiving text, voice notes, images, and attachments with a three-tier contact permission system (owner, trusted, untrusted). Unknown senders are quarantined until the owner approves them, blocking unsolicited prompt injection.
It triggers OpenClaw's wake hook on message arrival so the agent responds immediately instead of waiting for the next scheduled heartbeat.
When to use it
- Receiving task instructions from a trusted contact over Signal
- Sending voice replies to Signal messages using a local TTS engine
- Approving or rejecting new Signal contacts before the agent engages
- Transcribing incoming Signal voice notes for agent processing
- Triggering immediate agent wake when a new Signal message arrives
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: sends a voice note to the agent's Signal number
- 1signal-poll.sh receives the incoming voice message and writes it to the conversation log
- 2Poll script POSTs to OpenClaw /hooks/wake API to wake the agent immediately
- 3Agent reads pending_wakes and loads the sender's conversation history
- 4Agent transcribes the voice attachment using Whisper
- 5Agent generates a reply, converts it to m4a via ffmpeg, and sends it as a voice attachment via signal-send.sh
Agent sends a spoken reply back to the contact's Signal number within seconds of receiving the message
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.