@efengx/dragon-agent-executor

The autonomous edge execution engine for the Dragon Agent ecosystem.

Overview

@efengx/dragon-agent-executor is a lightweight, cloud-native agent executor designed to run inside restricted environments (like Docker containers). It serves as the primary bridge between the central Dragon Orchestrator and individual agents.

Its primary role is to "pull" deployment blueprints (n8n-style), execute the required setup steps autonomously, and report real-time telemetry back to the dashboard.

Key Features

• 🚀 Autonomous Provisioning: Agents bootstrap themselves by pulling blueprints from the orchestrator URL. No incoming SSH required.
• 🔄 Dynamic Placeholder Resolution: Automatically resolves {{ variable }} placeholders in blueprints using a global parameter context.
• 📡 Real-time Telemetry: Streams execution trajectories, logs, and status updates via a robust HTTP telemetry API.
• 🛠️ Self-Correction: Built-in "doctor" checks to ensure local Node.js and OpenClaw environments are healthy before starting.
• 📦 Configuration Sync: Automatically snapshots and reports local openclaw.json configurations to the cloud UI.

Usage (CLI)

Run the executor directly using npx:

npx -y @efengx/dragon-agent-executor@latest run \
  --agent-id "your-agent-id" \
  --orchestrator-url "https://dragon-orchestrator.ofengx.com"

Best Practices

1. Always use @latest: Ensure you are running the latest version to get the most up-to-date variable resolution and telemetry features.
2. Naming Blueprint Nodes: Use descriptive names like Cloud-NodeInstall or Gateway-Start to ensure clear visualization in the Dragon Web UI trajectory panel.
3. Variable Mapping: Define shared variables in a Global_Params node within your n8n workflow for clean, recursive resolution.

Environment Variables

• OPENCLAW_CONFIG_PATH: (Optional) Override the path to search for the OpenClaw configuration file.

License

MIT