← 返回 Skills 市场
Operator Hardening Pack
作者
Christopher Wheeler
· GitHub ↗
· v1.0.0
· MIT-0
67
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install operator-hardening-pack
功能描述
A unified operating workflow for hardening agent setups and running complex tasks reliably by combining bootstrap hardening, role orchestration, and handoff...
安全使用建议
This skill's goal (hardening agent workspaces) is reasonable, but the runbook tells the agent to read and modify internal memory/config files without declaring which paths or asking for explicit approvals. Before installing or running it: (1) ask the author for source code or a homepage and a clear list of exact files/paths the skill will access; (2) require explicit, per-action user approval before any reads/writes or external network calls; (3) test in an isolated/sandbox workspace first; (4) consider adding guards that prevent the skill from reading unrelated secrets or exporting data externally; and (5) refuse to install if the maintainer cannot justify the exact permissions and provide an auditable change log. These steps will reduce the risk that the skill inadvertently (or intentionally) reads or modifies sensitive agent state.
功能分析
Type: OpenClaw Skill
Name: operator-hardening-pack
Version: 1.0.0
The skill bundle consists entirely of Markdown documentation and metadata providing organizational guidelines for an AI agent's workflow. It focuses on 'hardening' agent configurations and managing role-based orchestration (e.g., Researcher, Builder, Editor) without any executable code, network requests, or instructions to access sensitive system files. The content in SKILL.md and references/pack-runbook.md promotes safety, approval policies, and structured handoffs.
能力评估
Purpose & Capability
The name/description (agent hardening, role orchestration, handoff) align with the runbook contents: auditing memory/files, trimming redundancy, establishing role contracts. However, the skill declares no required config paths, binaries, or credentials even though the runbook explicitly tells the agent to 'Check AGENTS/SOUL/USER/HEARTBEAT/memory files' and to make durable memory updates — an omission that makes the required access unclear.
Instruction Scope
SKILL.md and the runbook instruct the agent to inspect and modify agent memory/config files and to 'Deliver + log' and 'Durable memory updates'. Those are broad actions that can access sensitive data. The instructions do not specify which exact files/paths are safe to read or ask for explicit user approval before local reads/writes or any external transmission, creating scope creep and unclear boundaries.
Install Mechanism
This is an instruction-only skill with no install spec and no code files; nothing is written to disk by an installer. That minimizes supply-chain risk. However, the lack of code also means static scanners had no code to analyze, increasing uncertainty about runtime behavior.
Credentials
No environment variables, credentials, or config paths are declared, which superficially reduces credential risk. At the same time, the runbook expects access to 'AGENTS/SOUL/USER/HEARTBEAT/memory' which implies reading/writing internal agent state; because those paths are not declared, it's unclear what permissions the skill needs and whether it will access unrelated sensitive data.
Persistence & Privilege
The skill is not marked always:true and does not request autonomous privileges beyond the platform default. It does instruct making durable memory updates, which is normal for a hardening workflow but should be constrained by explicit user approval.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install operator-hardening-pack - 安装完成后，直接呼叫该 Skill 的名称或使用
/operator-hardening-pack触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release: unified hardening + orchestration + handoff workflow
元数据
常见问题
Operator Hardening Pack 是什么？
A unified operating workflow for hardening agent setups and running complex tasks reliably by combining bootstrap hardening, role orchestration, and handoff... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 67 次。
如何安装 Operator Hardening Pack？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install operator-hardening-pack」即可一键安装，无需额外配置。
Operator Hardening Pack 是免费的吗？
是的，Operator Hardening Pack 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Operator Hardening Pack 支持哪些平台？
Operator Hardening Pack 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Operator Hardening Pack？
由 Christopher Wheeler（@cwheeler67）开发并维护，当前版本 v1.0.0。
推荐 Skills