toq

What This Skill Does

The toq skill enables secure, agent-to-agent communication within the OpenClaw ecosystem using the toq protocol. It acts as a bridge for distributed AI agents to exchange data, coordinate tasks, and facilitate collaborative problem-solving. By setting up an agent endpoint, you can manage secure connections, approve or revoke access for other agents, and handle message routing across network boundaries. The skill manages the lifecycle of your agent’s identity, providing tools to verify connectivity, configure DNS-based discovery, and audit incoming communication requests.

Installation

To install this skill, use the clawhub command: clawhub install openclaw/skills/skills/anshuldesai/toq

Once installed, initialize your node by running the configuration setup. Note that toq is currently in alpha stage—while it is powerful for distributed agent research and communication, it should not be used to transmit sensitive, personal, or classified data. Always verify the status of your agent with toq doctor after installation to ensure all dependencies and networking paths are correctly configured.

Use Cases

• Decentralized Coordination: Allow multiple agents to work on a single complex task by delegating sub-tasks through direct messaging.
• Agent Networks: Build a private mesh of trusted agents that can share information or provide services to each other without a central server.
• Secure Communication: Implement verified communication channels where both parties must explicitly approve the connection, preventing unauthorized access.
• Distributed Discovery: Use DNS discovery to maintain persistent communication links with other agents even when IP addresses change.

Example Prompts

1. "Setup my agent for toq communication, I want to call my agent 'research-bot'."
2. "Can you check my current toq connection status and see if there are any pending approval requests?"
3. "Send a secure ping to toq://central-hub/coordinator and ask for the latest project task list."

Tips & Limitations

• Alpha Warning: The protocol is experimental. Performance and stability may vary depending on network topology.
• Bidirectional Trust: Remember that connections are mutual. Even if you approve someone, they must also approve your requests before message flow starts.
• Security First: Regularly check your toq permissions to ensure your allowlist remains clean. Use patterns like toq://host/* carefully to avoid over-privileged access. If you suspect an agent is acting maliciously, use toq block immediately to terminate all ongoing threads and prevent future communication.