clawzembic

What This Skill Does

Clawzembic is a high-precision performance audit tool designed specifically for OpenClaw instances. It functions as a diagnostic engine that evaluates your environment against six performance-critical categories: context injection efficiency, cron job health, session bloat, configuration optimization, active skill overhead, and transcript size management. By simulating a 'Lighthouse' style audit, it identifies exactly where your tokens are being wasted, which background tasks are misconfigured, and how your current model utilization compares to best practices. Because it utilizes only the Python standard library, it ensures zero-dependency reliability and maximum security across any environment.

Installation

To integrate Clawzembic into your OpenClaw environment, ensure you have the clawhub utility installed. Run the following command in your terminal:

clawhub install openclaw/skills/skills/anna-claudette/clawzembic

Once installed, you can invoke the auditor directly via bash skills/clawzembic/lean-audit.sh. No additional packages are required, and the script is designed to run natively on any machine with Python 3.8 or higher.

Use Cases

Clawzembic is essential for power users managing high-volume OpenClaw instances. It is particularly valuable when you notice latency in response times or unexpected token consumption spikes. Use it to audit remote servers before a big project to ensure your configuration is lean, to perform routine maintenance on cron jobs that may be silently degrading performance, or to generate quick diagnostic reports for troubleshooting session bloat. It provides both human-readable text output for quick review and JSON output for automated dashboard monitoring or CI/CD integration.

Example Prompts

1. "Run a full Clawzembic audit on the current instance and summarize the top three areas where I can save tokens."
2. "My instance is feeling slow. Can you run the audit in --fix mode and tell me what configuration changes you suggest to improve my grade?"
3. "Check the performance of my remote VM at user@claudette using the clawzembic skill and give me a letter grade for its current health."

Tips & Limitations

To get the most accurate results, always ensure you have SSH key-based authentication enabled for remote auditing. While the tool is highly effective at identifying bloat, it cannot modify your core system files unless the --fix flag is explicitly provided, making it safe for read-only diagnostics. Remember that the grade scale is subjective to the specific use case; an 'F' isn't always a catastrophe, but it indicates significant room for model right-sizing and context cleanup. Always review the suggested fixes before applying them to mission-critical production instances.