Cross-Agent Collaboration for Openclaw

A specialized tool for distributing AI tasks and managing file transfers across multiple machines in a local network using SSH.

yuleleyo
v1.0.0
Mar 5, 2026
0
1.1k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install skill-cross-agent

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install skill-cross-agent using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is Cross-Agent Collaboration?

The Cross-Agent skill enables seamless collaboration between multiple instances of AI agents within a local area network (LAN). By leveraging SSH protocols, this skill allows one agent to act as a controller, scanning for available peers, testing connections, and delegating complex instructions to remote Openclaw Skills setups. It effectively turns a single machine into a coordinator for a decentralized network of AI workers.

Whether you need to offload heavy processing to a secondary machine or synchronize data between distributed agents, Cross-Agent provides the necessary CLI tools for remote execution and file management. It is designed for developers who require a scalable way to manage multi-agent workflows without complex cloud infrastructure, relying instead on the proven reliability of SSH and Openclaw Skills.

Cross-Agent Collaboration Use Cases

  • Distributing heavy computational tasks to a more powerful machine in the LAN.
  • Synchronizing research documents and generated code between multiple AI workstations.
  • Monitoring active agent sessions on remote hardware in real-time.
  • Executing shell commands across a cluster of local agents for maintenance or data gathering.
  • Automating multi-step workflows that require local file access on different physical devices.

How Cross-Agent Collaboration Works

  1. The controller agent scans the local network segment using the scan command to identify online Openclaw Skills instances.
  2. Connection parameters like IP, username, and password are configured or tested using the test and config commands.
  3. The controller sends specific task instructions to a target agent using the send command.
  4. Remote agents process the tasks and store results locally or in active sessions.
  5. The controller retrieves the output files or executes follow-up commands via scp-based get/put operations or direct exec instructions.

Cross-Agent Collaboration Setup

Ensure your system meets the requirements by installing the necessary networking tools:

sudo apt install sshpass ssh scp ping nc

The target machine must have SSH enabled (Port 22), the OpenClaw Gateway running, and be accessible via the LAN. Initialize the connection on your primary machine:

openclaw cross-agent config --default-user <username> --default-pass <password> --default-ip <target-ip>
openclaw cross-agent test

Cross-Agent Collaboration Data Schema & Taxonomy

Cross-Agent manages data primarily through remote session tracking and file transfers. It organizes metadata as follows:

Category Description
Connection Config Stores default IP, User, and Password for persistent Openclaw Skills sessions.
Session Metadata Tracks active OpenClaw sessions on remote machines accessible via the sessions command.
File Transfer Uses standard Unix paths for get and put operations between local and remote directories.
Task Instructions Command strings sent to remote agents to trigger specific agentic workflows.

Cross-Agent Collaboration Advanced Features

  • Interactive Wizard: Use the openclaw cross-agent wizard for a guided setup of multi-machine collaboration.
  • Remote Command Execution: Run arbitrary shell commands on target machines using the exec functionality.
  • Network Discovery: Automated LAN scanning to find peer Openclaw Skills instances without manual IP entry.
  • Session Inspection: Peek into the current state of remote agent tasks to monitor progress across the network.
  • Automated Parameter Injection: Configure default credentials to streamline command usage and reduce repetitive typing in scripts.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*