Agent优化专家

This Skill addresses common operational challenges in AI Agent runtime, such as frequent Cron job failures, tool invocation errors, workflow interruptions, and performance degradation. Developers often need to manually check logs, inspect configurations, and restart services—a tedious and error-prone process. Agent Optimization Expert automates the entire closed loop of diagnosis, repair, verification, and learning, significantly reducing the need for manual intervention.

Usage is straightforward: simply describe the issue in natural language, e.g., "system diagnosis," "fix cron failure," or "check system health." It will automatically execute the diagnostic process: first scanning Cron job status, tool error logs, and system resource usage, then applying fixes based on built-in repair templates (e.g., fix-templates.md), and finally verifying results and logging errors. It also supports automatic adaptation to dual environments (OpenClaw and Hermes) without requiring manual command switching.

It is ideal for teams or individuals running AI Agents long-term, especially developers and operations engineers who have deployed OpenClaw or Hermes Agents. Whether for routine inspection, failure recovery, or performance optimization, it significantly enhances Agent stability. For users with extensive Cron jobs and third-party tool integrations, this Skill is almost essential.

After initial deployment, it is recommended to manually trigger a full inspection to ensure all dependent services (e.g., SearXNG, Firecrawl) are healthy. Note that it primarily addresses Agent execution-level issues and is not suitable for business logic errors or data query tasks. Additionally, L2/L3 operations (e.g., modifying core configurations, deleting data) will prompt user confirmation to prevent accidental actions.