Agent优化专家

This Skill addresses common pain points in the daily operation of AI Agents, such as recurring Cron task failures, tool invocation errors, workflow interruptions, and performance degradation. In practice, developers and DevOps engineers often need to manually check logs, switch platform commands, and consult repair templates, which is inefficient and prone to oversight. This Skill automates the diagnosis → repair → verification → learning loop, significantly reducing Agent maintenance overhead.

Usage is straightforward: simply describe the issue in natural language, e.g., "Cron task failed" or "Check system health." The Skill will automatically execute the diagnostic process: first listing all Cron tasks or tool invocation statuses, pinpointing the cause of failure, applying minimal changes based on built-in repair templates (references/fix-templates.md), and finally verifying the repair results. It is compatible with both OpenClaw and Hermes Agent environments, automatically detecting the current platform and switching to the corresponding commands without manual configuration.

It is ideal for DevOps engineers, Agent developers, and full-stack developers who frequently manage Agent instances. Particularly for teams already using OpenClaw or Hermes Agent to build automated workflows, this Skill can facilitate daily inspections, self-healing repairs, and continuous optimization. It is also suitable for individual developers maintaining the stable operation of personal Agents.

It is recommended to prioritize using this Skill for diagnosis whenever an Agent anomaly occurs, rather than making manual modifications directly. Note that it primarily targets technical issues at the Agent execution level and is not suitable for business logic troubleshooting, Lark content editing, or data query tasks. The Skill includes a tiered security mechanism (L0-L3) for operations; actions involving modifications to core configurations or data deletion will require user confirmation to prevent unintended changes.