Wreckit Skill

What This Skill Does

The Wreckit Skill (jmanhype fork) is a powerful, autonomous software factory designed to integrate the Clawdbot with high-concurrency, isolated software development workflows. Unlike standard AI loops, this skill leverages a unique "Cattle vs. Pets" architecture. By default, it operates in "Cattle Mode," utilizing ephemeral Firecracker microVMs (Sprites) to execute code in complete isolation, ensuring that every task starts from a clean slate. This eliminates state contamination and significantly enhances system security. The skill acts as an autonomous engineer, capable of ideating new technical tasks, executing development loops, self-healing through diagnostics, and even summarizing progress via visualization videos. It is the ultimate tool for developers looking to automate the entire lifecycle of a software project.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/jmanhype/wreckit-skill Ensure you have reviewed the setup guide in SETUP.md to properly configure the jmanhype-forked Wreckit CLI before attempting to run your first task.

Use Cases

• Automated Feature Development: Trigger the Wreckit loop on specific item IDs to have the agent write, test, and commit code autonomously.
• Self-Healing Infrastructure: Use the doctor command to allow the agent to perform diagnostics on the current environment and automatically repair issues.
• Autonomous Ideation: Use the dream command to let the agent parse your project structure, identify technical debt, and generate a self-managed roadmap.
• Knowledge Harvesting: Use the learn command to analyze completed tasks and transform those successful workflows into persistent, reusable agent skills.
• Reporting: Generate quick visualization videos of development progress for stakeholder reviews using the summarize command.

Example Prompts

1. "Wreckit run 096"
2. "Wreckit dream"
3. "Wreckit learn all"

Tips & Limitations

• Security: The default Cattle Mode is highly recommended for all operations. Use Pet Mode only for rapid, local debugging where host-level access is explicitly required and monitored.
• Dependency Management: When running individual tasks, ensure all prerequisites for that specific item ID are met, as the agent relies on a consistent internal state for dependency satisfaction.
• Performance: While Pet Mode is faster, it poses a risk of state contamination; always revert to Cattle Mode once debugging is finished.