Agent优化专家

This Skill addresses common operational pain points in AI Agent production environments, such as frequent Cron job failures, tool invocation exceptions, workflow interruptions, and performance degradation. In practice, developers often spend significant time manually checking logs, identifying root causes, and executing fixes—a tedious process prone to oversight. This Skill automates the entire cycle, freeing developers from repetitive troubleshooting so they can focus on core business logic.

Usage is straightforward: simply install and enable this Skill in your Agent environment, then use natural language commands to trigger system diagnostics, fix Cron issues, or check health status. Its core workflow follows a closed loop of "Detection → Diagnosis → Repair → Verification → Learning → Evolution." It automatically identifies the current runtime platform (OpenClaw or Hermes) and switches to the corresponding repair commands. For example, when a Cron job fails, it will automatically list task status, pinpoint errors, and execute fixes based on repair templates.

It is ideal for operations engineers, Agent developers, and project managers responsible for maintaining the stability of AI Agent systems—especially teams already using the OpenClaw or Hermes Agent frameworks, or those facing complex production environments with multi‑agent coordination, scheduled tasks, and extensive toolchain dependencies. For teams pursuing high system availability and self‑healing capabilities, this is an indispensable operational tool.

We recommend configuring a scheduled self‑check mechanism (e.g., via Cron Job or heartbeat file) immediately after deploying the Agent to production. Note that this Skill primarily handles generic platform‑ and tool‑layer failures (such as permission, configuration, or resource issues) and is not intended for debugging specific business logic errors or performing application‑layer tasks like editing Feishu content. Please be clear about its scope when using it.