Menu
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
npx skills add https://github.com/openocta/openocta_skills --skill server-patrolThe command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
Showing SKILL.md
| name | server-patrol |
| description | Linux 与网络设备只读健康巡检；直接运行 scripts/patrol.sh，配置由平台侧信道注入，Agent 不读取环境变量 |
面向 Linux 服务器 与 网络设备（路由器、交换机、防火墙等）的只读巡检 Skill。
| 对象 | 检查项 |
|---|---|
| Linux | uptime、负载、CPU、内存、磁盘、systemd 服务、Docker |
| 网络设备 | Ping 连通、TCP 端口、SNMP（可选）、SSH 只读命令（Cisco / 华为 / H3C） |
巡检目标与凭证由 OpenOcta 平台在运行 patrol.sh 前注入进程环境（来自 Skill 包内 config.env + Config Vault），不对大模型暴露。
env、printenv、echo $PATROL_*、grep PATROLcat/read_file 读取 config.envchmod +x scripts/patrol.sh
./scripts/patrol.sh --list # 查看已配置目标（无目标时再报错）
./scripts/patrol.sh --all # 批量 Linux + 网络
./scripts/patrol.sh --servers # 仅 Linux
./scripts/patrol.sh --network # 仅网络设备
./scripts/patrol.sh --network --tag core
./scripts/patrol.sh --all --format json
--list 为空或脚本报「未配置目标」，告知用户：需在控制台 Config Vault / Skill config.env 中配置巡检目标，不要自行构造 export PATROL_*。reports/ 下报告解读结果，按下方「处理思路」给出建议。./scripts/patrol.sh --list 或 ./scripts/patrol.sh --all| 发现 | 建议 |
|---|---|
| Ping CRIT | 查链路、ARP、ACL、设备电源 |
| 端口 closed | 确认服务监听、防火墙、安全组 |
| Cisco/华为 CPU 高 | 查会话数、路由震荡、广播风暴 |
| 接口 down | 查物理链路、光模块、对端端口 |
| Linux 磁盘/内存 CRIT | 同服务器巡检常规处理 |
| SSH/SNMP 失败 | 查凭证、SNMP ACL、管理 VLAN（勿向用户索要明文密码） |
PATROL_* 列表、Config Vault 引用）见 README.md 与 config.env，Agent 无需阅读Slow query diagnosis with root cause analysis and index optimization recommendations. Identifies TOP N slow queries, analyzes scan efficiency, calculates index hit rate, and suggests covering indexes — turning a 20-minute manual diagnosis into 8 seconds.
Replication lag root cause analysis with automated resolution workflow. Detects lag, identifies blocking queries on the master, presents KILL candidates with safety classification, and executes KILL only after double user confirmation. Reduces recovery from 20 minutes to 2 minutes.
Full-chain MySQL deadlock analysis. Captures InnoDB status, parses lock chains, identifies deadlock patterns (AB-BA / Gap Lock / FK), finds current lock waits and long transactions, outputs structured report with fix recommendations. Supports MySQL 5.7+ performance_schema and 5.6 compatibility mode.
Based on SOC occupation classification