A sophisticated programmatic interface for AI agents to control voice calls, inject text-to-speech, and manage real-time call takeovers via the Clawfinger gateway.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawfinger
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install clawfinger using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Clawfinger is a specialized plugin for Openclaw Skills that establishes a persistent bridge between AI coding agents and the Clawfinger voice gateway. It enables developers to integrate cellular voice capabilities directly into their agentic workflows, allowing agents to act as real-time voice assistants. By connecting to an Android device via ADB and the PhoneBridge app, this skill provides the necessary tools to bridge the gap between Large Language Models and traditional telephony.
Through this integration, agents can monitor active call states, manage conversation history, and dynamically inject context to ensure the AI has the most relevant facts during a live discussion. It is a critical tool for developers building automated voice support systems, outbound notification agents, or interactive voice response (IVR) replacements that require the nuanced intelligence of modern AI models.
clawfinger_dial tool to initiate a call or waits for the gateway to report an incoming session.clawfinger_takeover to intercept user input and provide custom responses instead of the gateway's default logic.clawfinger_turn_wait to receive live transcripts and clawfinger_turn_reply to speak back to the caller.clawfinger_hangup or releases the session back to the gateway's automated handling.First, ensure the Clawfinger gateway is active and accessible. Install the required dependencies in the plugin directory:
npm install @sinclair/typebox
Configure the skill in your ~/.openclaw/openclaw.json file by adding the plugin path and gateway credentials:
{
"plugins": {
"load": { "paths": ["/path/to/gateway/openclaw/clawfinger"] },
"entries": {
"clawfinger": {
"enabled": true,
"config": {
"gatewayUrl": "http://127.0.0.1:8996",
"bearerToken": "your_token_here"
}
}
}
}
}
The skill manages call data through several structured components:
| Data Component | Description |
|---|---|
| Session State | Contains the unique Session ID (SID), conversation history, and active status. |
| Call Policy | Defines global behaviors like auto_answer, greetings, and max_duration. |
| Instruction Set | Categorized into Global, Session, or Turn-based prompts to guide LLM behavior. |
| Knowledge Context | Injected facts or documents that the voice assistant uses to provide accurate answers. |
| Takeover Metadata | Tracks request IDs for specific conversation turns during manual agent control. |
Loading
A protocol-level skill for hijacking live voice gateway turns to provide custom agent responses or human intervention.

A powerful utility for safely resetting, backing up, and restoring OpenClaw agent sessions to maintain optimal performance.

A CLI tool for deploying applications to secure, private Fly.io networks reachable only via Tailscale.

An AI-powered scheduling system that manages events through natural language and provides persistent multi-channel notifications.

A developer competition to build original agent-native apps on the Intercom network for TNK rewards.

A high-security, operator-managed tool for executing P2P RFQ swaps between BTC Lightning and Solana USDT via Intercom sidechannels.








































