What it does
OpenClaw operations skill with health checks, repair scripts, watchdogs, update triage, and security scans.
- Connects Instruction files the AI reads
- Account or key Not confirmed yet (README not read yet)
- Category 自动化与智能体
给作者:在 README 中加徽章
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/cathrynlavery-openclaw-ops)相似的 Skill
同类工作,排序与 Skill 列表相同
Skill 和 MCP 有什么不同
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
适用的应用
Codex
依据开发者 README 和 repo 中的文件夹
安装
Is it safe?
- Who made it
GitHub account cathrynlavery (person), 7 years old
- Use and upkeep
★ 233 stars, few users so far · code last changed 2026年7月4日
- Any scripts?
28 scripts that can run on a computer: scripts/agent-dirs-audit.sh, scripts/backup-rotate.sh, scripts/check-update.sh, scripts/codex-perf-check.sh and 24 more
- Uses the internet?
Scripts that go online: scripts/health-check.sh
- Changes machine-wide settings?
No hooks or writes to machine settings found
Checked automatically from the repo files at commit b6dc8cf (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Option 1Upload to claude.ai (web and app)
Uploading to claude.ai would fail: the description is over 200 characters (222). Use "Let the AI install it" instead
Option 1让 Claude Code 帮你安装
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /openclaw-ops,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
打开终端
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
把 skill 文件夹复制到 ~/.claude/skills
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
重新打开 Claude Code,输入 /openclaw-ops,或直接提出相关任务,Claude 会自动选用
ChatGPT 桌面应用
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
Option 1让 ChatGPT 帮你安装
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Option 2自己放入文件夹
把 openclaw-ops 文件夹放到 ~/.agents/skills/openclaw-ops
~ 是你的主文件夹($HOME)
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
打开终端
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
把 skill 文件夹复制到 ~/.agents/skills
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
安装之后
- ChatGPT 桌面应用:打开侧边栏中的 Skills,确认 Skill 已出现
- 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
- Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
网页版和手机版 ChatGPT
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Option 1让 Codex 帮你安装
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
打开终端
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
把 skill 文件夹复制到 ~/.agents/skills
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
重新打开 Codex 并提出相关任务,Codex 会列出这个 skill
Option 1让 Cursor 帮你安装
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 openclaw-ops
Checked against official docs 2026年11月12日 · Source: cursor.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
打开终端
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
把 skill 文件夹复制到 ~/.cursor/skills
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
在 Cursor 的 Agent 聊天中输入 / 并选择 openclaw-ops
Option 1Let the AI install it
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
For Terminal users
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
查看原始 SKILL.md
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
显示前 8,000 个字符 · 在 GitHub 打开完整文件
---
name: openclaw-ops
description: Use when installing, configuring, troubleshooting, securing, or performing a health check on OpenClaw gateway setups — including channel integrations, exec approvals, cron jobs, agent sessions, and operational maintenance.
---
# OpenClaw Ops
You are an expert OpenClaw administrator. Use the scripts below to diagnose and fix issues — they contain the implementation logic. Reach for scripts first; only write manual steps when no script covers the case.
## Reference Documentation
- [cli-reference.md](docs/cli-reference.md) — Complete CLI command reference
- [troubleshooting.md](docs/troubleshooting.md) — Common issues and solutions
- [channel-setup.md](docs/channel-setup.md) — Platform-specific setup guides
- [security-guide.md](docs/security-guide.md) — Active security defense guide
- [docs.openclaw.ai](https://docs.openclaw.ai) — Official documentation
---
## Scripts
All scripts live in `scripts/` relative to this skill (typically `~/.openclaw/skills/openclaw-ops/scripts/`). Always use that full path when suggesting commands to users.
On Knox's machine, the canonical ops checkout is `/Users/knox/Developer/openclaw-ops`. Prefer those scripts over the installed skill snapshot under `~/.agents/skills/openclaw-ops`, which can lag behind. If the two differ, treat the Developer checkout as authoritative and sync/reinstall the skill snapshot as follow-up work.
| Script | When to use |
|--------|-------------|
| `heal.sh` | First thing on any health check — fixes gateway, auth mode, exec approvals, crons, and stuck sessions in one pass |
| `post-update.sh` | Run after `openclaw update` — orchestrates check-update, heal, workspace reconcile, security scan, and final health check in sequence |
| `watchdog.sh` | Continuous monitoring; run every 5 min via LaunchAgent. HTTP health check → auto-restart → escalation after 3 failures |
| `watchdog-install.sh` | Set up the watchdog as a macOS LaunchAgent (survives reboots) |
| `watchdog-uninstall.sh` | Remove the LaunchAgent |
| `check-update.sh` | After a version change — detects breaking config changes, explains them; `--fix` to auto-repair |
| `health-check.sh` | URL/process health checks for gateway-adjacent services; copy `templates/health-targets.conf.example` first |
| `log-sweep.sh` | Mandatory when the user says “check the logs” or quotes an error — searches gateway logs plus `/tmp/openclaw/openclaw-*.log` / `/private/tmp/openclaw/openclaw-*.log`; use the user’s exact error string first, then the default broad pattern. |
| `session-monitor.sh` | Agent is alive but misbehaving — retry loops, hangs, auth loops, noisy failures |
| `session-search.sh` | Search session history by keyword; redacts secrets by default |
| `session-resume.sh` | Build a readable markdown resume for a single session (compaction-first, then point-of-failure) |
| `prompt-truncation-report.sh` | Report bootstrap truncation warnings from the latest session per agent. Use when users say “prompt too long,” “instructions too long,” or the bootstrap context looks incomplete. |
| `cron-optimize.sh` | Audit agent cron jobs for missing `--light-context`; `--fix` enables it and adds a default thinking level only when one is not already set. |
| `cron-error-inspector.sh` | Format erroring cron jobs from cron state, including last error, reason, consecutive count, last-run age, and a truncated payload preview. |
| `remediation-board.sh` | Human/agent repair board for surfaced ops findings. Import cron or machine incidents, track recurring bugs, hacks/workarounds, upstream watches, incident notes, hypotheses, steps tried, and verification state. |
| `agent-dirs-audit.sh` | Audit unconfigured dirs under `~/.openclaw/agents/`. Default is dry-run; `--archive` moves dormant dirs to `_archived/YYYY-MM-DD/`, `--delete-empty` removes empty dirs. |
| `backup-rotate.sh` | Rotate generic `*.bak*` files across `~/.openclaw`, grouped by the path prefix before `.bak`. Keeps the newest N per group; dry-run by default, `--apply` to delete. |
| `context-audit.sh` | Audit AGENTS.md, MEMORY.md, and SOUL*.md for file bloat. Reports path, token estimate (chars/4), and mtime, ranked largest-first above a token threshold. |
| `session-purge.sh` | Reclaim disk + cut session context bloat. Purges stale session index entries, orphan cron/subagent sessions, old `.bak` files, and orphan `.jsonl` transcripts. Dry-run by default; `--apply` to execute. |
| `workspace-auto-commit.sh` | Commit dirty OpenClaw workspace repos locally. Defaults to `~/.openclaw/workspace`; use `--workspace PATH` for an agent repo or `--all` for every `workspace*` repo. Never pushes. |
| `workspace-git-audit.sh` | Audit `~/.openclaw/workspace*` repos for git status and auto-commit cron coverage. Use `--show-cron` to print suggested cron add commands for uncovered repos; `--strict` fails on uncovered or dirty repos. |
| `daily-digest.sh` | Incident, activity, watchdog, and cost summary for the last N hours |
| `incident-manager.sh` | Sourced helper for incident lifecycle (used by session-monitor and other scripts) |
| `skill-audit.sh` | Before `clawhub install` — scan skill for secrets, injection, dangerous commands; outputs LOW/MEDIUM/HIGH risk score |
| `security-scan.sh` | Config hardening compliance check (0-100); `--fix` for auto-repair; `--drift` for file change detection; `--credentials` to scan for leaked secrets; `--include-sessions` includes bulky logs/session/runtime files normally skipped |
| `codex-perf-check.sh` | Check/fix four GPT-5.x performance opt-ins (strict execution, personality overlay, thinking level, Codex harness). Requires v2026.4.x+. `--fix` to apply. |
### Quick start examples
```bash
# One-pass heal:
bash scripts/heal.sh
# Install always-on watchdog (macOS):
bash scripts/watchdog-install.sh
# Check GPT-5.x agent performance settings:
bash scripts/codex-perf-check.sh
bash scripts/codex-perf-check.sh --fix # apply fixes
# Run behavioral session monitoring:
bash scripts/session-monitor.sh --verbose
# Search sessions for auth failures:
bash scripts/session-search.sh "unauthorized" --limit 10
# Build a resume for one session:
bash scripts/session-resume.sh ~/.openclaw/agents/knox/sessions/<session>.jsonl
# Check bootstrap truncation warnings:
bash scripts/prompt-truncation-report.sh
bash scripts/prompt-truncation-report.sh --agent atlas --json
# Audit cron jobs for missing light-context:
bash scripts/cron-optimize.sh
bash scripts/cron-optimize.sh --fix --level low
# Inspect cron failures:
bash scripts/cron-error-inspector.sh
bash scripts/cron-error-inspector.sh --agent atlas --consecutive 2
# Track surfaced findings through completion:
bash scripts/remediation-board.sh import-cron-errors
bash scripts/remediation-board.sh import-incidents
bash scripts/remediation-board.sh list
bash scripts/remediation-board.sh set cron:<job-id> fixed-awaiting-rerun --note "payload fixed"
# Mandatory: when investigation finds a real local OpenClaw error,
# regression, hack/workaround, security concern, or recurring ops finding,
# create/update a board item immediately. The board is the local repair loop;
# upstream links are optional metadata only when an external fix also exists.
bash scripts/remediation-board.sh add-incident log-gap "Log sweep missed runtime errors" --evidence "tmp OpenClaw log contained active failure"
bash scripts/remediation-board.sh close-criteria log-gap "Local log-sweep catches the failure class and the installed workflow is synced"
# Track a recurring bug / incident note (check existing board first):
bash scripts/remediation-board.sh list --type incident
bash scripts/remediation-board.sh show telegram-split
bash scripts/remediation-board.sh add-incident telegram-split "Telegram topic replies split" --evidence "Observed in forum topic"
bash scripts/remediation-board.sh hypothesis telegram-split "Preview draft lane sends instead of edits" --confidence medium
bash scripts/remediation-board.sh tried telegram-split --讨论
提问、分享技巧或报告无法使用。按“有帮助”数量排序。
来自 GitHub
- GitHub 星标
- 233
- 开发者
- cathrynlavery
- 许可证
- MIT
- 最近代码更新
- 2026年7月4日
- repo 中的 SKILL.md 文件(含译本和模板)
- 1
- 主题
- automation, openclaw, shell-scripts, skills