什么是 Agent GitOps？

Agent GitOps 为管理 Openclaw 技能提供专业级的 DevOps 工作流。由于 AI 智能体会频繁更新或修改自身能力，引入破坏性变更的风险随之增加。该技能通过实施基于 Git 的追踪系统来降低此类风险，记录每一次状态变更，从而实现智能体逻辑的无缝回滚和稳定版本化。它确保对 Openclaw 技能的每次修改都是可追踪、可逆且可验证的。

下载入口:https://github.com/openclaw/skills/tree/main/skills/trypto1019/arc-skill-gitops

安装与下载

1. ClawHub CLI

从源直接安装技能的最快方式。

2. 手动安装

将技能文件夹复制到以下位置之一

优先级：工作区 > 本地 > 内置

3. 提示词安装

将此提示词复制到 OpenClaw 即可自动安装。

Agent GitOps 应用场景

• 为快速演进的 AI 智能体能力提供版本控制
• 针对破坏性技能更新或配置错误实现快速恢复
• 审计 Openclaw 技能随时间推移而发生的变更
• 为本地智能体开发和测试实施类似 CI/CD 的门禁机制

1. init 命令在技能目录中初始化本地仓库，开始追踪代码库。
2. 开发人员或智能体创建快照，以标记工作流当前的稳定状态。
3. deploy 命令在应用更新前自动执行快照流程，确保存在回退方案。
4. 用户可以列出所有变更的历史记录，以识别特定的还原点。
5. rollback 命令利用 Git checkout 逻辑，将技能立即恢复到先前的快照状态。

Agent GitOps 配置指南

要开始为您的 Openclaw 技能使用 Agent GitOps，请确保系统中已安装 python3 和 git。运行以下命令：

# 为特定技能初始化追踪
python3 {baseDir}/scripts/gitops.py init --skill ~/.openclaw/skills/my-skill/

# 在重大变更前对状态进行快照
python3 {baseDir}/scripts/gitops.py snapshot --skill ~/.openclaw/skills/my-skill/ --tag "stable-v1"

# 检查所有受追踪技能的状态
python3 {baseDir}/scripts/gitops.py status

Agent GitOps 数据架构与分类体系

Agent GitOps 通过本地 git 仓库和集中式清单的结合来管理数据：

name: arc-skill-gitops
description: Automated deployment, rollback, and version management for agent workflows and skills.
user-invocable: true
metadata: {"openclaw": {"emoji": "??", "os": ["darwin", "linux"], "requires": {"bins": ["python3", "git"]}}}

Agent GitOps

Automated deployment and rollback for agent workflows. Track skill versions, deploy with confidence, roll back when things break.

Why This Exists

Agents install and update skills constantly. When an update breaks something, you need to:

1. Know what changed
2. Roll back fast
3. Track which version was stable

This skill manages that lifecycle.

Commands

Initialize tracking for a skill

python3 {baseDir}/scripts/gitops.py init --skill ~/.openclaw/skills/my-skill/

Snapshot the current state (before updating)

python3 {baseDir}/scripts/gitops.py snapshot --skill ~/.openclaw/skills/my-skill/ --tag "pre-update"

Deploy a skill update (snapshots current state first)

python3 {baseDir}/scripts/gitops.py deploy --skill ~/.openclaw/skills/my-skill/ --tag "v1.1"

List all snapshots for a skill

python3 {baseDir}/scripts/gitops.py history --skill ~/.openclaw/skills/my-skill/

Roll back to a previous snapshot

python3 {baseDir}/scripts/gitops.py rollback --skill ~/.openclaw/skills/my-skill/ --tag "pre-update"

Check status of all tracked skills

python3 {baseDir}/scripts/gitops.py status

Run pre-deploy checks (integrates with arc-skill-scanner if available)

python3 {baseDir}/scripts/gitops.py check --skill ~/.openclaw/skills/my-skill/

Data Storage

Snapshots and metadata are stored in ~/.openclaw/gitops/.

How It Works

1. init creates a git repo in the skill directory (if not already one) and records it in the tracking manifest
2. snapshot commits the current state with a tag
3. deploy takes a pre-deploy snapshot, then commits the new state
4. rollback restores a tagged snapshot using git checkout
5. check runs arc-skill-scanner (if installed) as a pre-deploy gate
6. history shows all tagged snapshots with timestamps
7. status shows all tracked skills and their current/latest versions

Tips

• Always snapshot before updating a skill manually
• Use deploy instead of manual updates — it snapshots automatically
• Combine with arc-skill-scanner for security-gated deployments
• Tags should be descriptive: "v1.2", "pre-security-patch", "stable-2026-02-15"