antigravity-swarm

What This Skill Does

The antigravity-swarm skill is a powerful orchestration framework for OpenClaw that enables you to deploy autonomous sub-agents within your local development environment. By leveraging a centralized Manager (planner.py) and an Orchestrator (orchestrator.py), the skill can decompose complex high-level project goals into actionable tasks, assign them to sub-agents, and track their progress through a visual interface. It acts as a force multiplier for developers, allowing you to offload repetitive or multi-step tasks like test generation, refactoring, or documentation synthesis to background processes.

Installation

To integrate this skill into your OpenClaw environment, use the internal installer command: clawhub install openclaw/skills/skills/0xnagato/antigravity-swarm

Use Cases

This skill is ideal for complex, multi-step development workflows. Use it when:

• You need to perform large-scale refactoring across multiple files where task separation prevents conflicts.
• You are building a new feature and require a diverse set of sub-agents to handle specific concerns (e.g., writing tests, updating READMEs, and implementing logic concurrently).
• You have parallelizable research tasks, such as analyzing a large directory structure and summarizing findings simultaneously.

Example Prompts

1. "Use antigravity-swarm to build a Python-based Snake game. Let the agent team plan the structure and execute the file creation."
2. "Dispatch a sub-agent to generate comprehensive unit tests for the current file I am working on."
3. "Run a mission to scan this directory and create a markdown file summarizing all current technical debt in the project."

Tips & Limitations

• Safety First: The system utilizes a 'Plan Mode' by default. Always review the generated subagents.yaml before proceeding to execution to prevent unintended file modifications.
• Avoid Manual Editing: Never edit task_plan.md, findings.md, or subagents.yaml while the Orchestrator is actively running. This creates race conditions that can corrupt the agent mission status.
• Visibility: Use the terminal TUI (orchestrator.py) for the most comprehensive view of active sub-agent output. If running in an IDE Agent context, rely on JSON logs for real-time monitoring.
• Resource Management: Since sub-agents are independent processes, ensure you do not spawn more agents than your local system resources can handle to avoid memory exhaustion.