← 返回 Skills 市场
anwhere

OpenClaw Agent Clone

作者 Anwhere · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
291
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install claw-clone
功能描述
一键导出或导入小龙虾配置,支持配置共享与克隆,自动备份并过滤敏感信息,方便技能安装与应用恢复。
使用说明 (SKILL.md)

claw-clone - OpenClaw Agent Clone Skill

🦞 One-click export/import of OpenClaw agent configuration for sharing and cloning.

Triggers

  • "export" / "export config" / "export package" / "clone yourself"
  • "import" / "import config" / "import package" / "clone"
  • When user wants to share or obtain agent configuration

Features

Export Mode (claw-clone-out)

Export all key information of the current agent into a shareable package:

  1. Identity Files

    • IDENTITY.md - Identity definition
    • SOUL.md - Soul/personality settings
    • USER.md - User information
  2. Configuration Files

    • AGENTS.md - Agent rules
    • HEARTBEAT.md - Heartbeat configuration
    • TOOLS.md - Tools configuration
  3. Skills Information

    • List of installed skills with versions
  4. OpenClaw Configuration (sensitive info filtered)

  5. Optional: Memory

    • Ask user whether to include MEMORY.md

Import Mode (claw-clone-in)

Clone another agent's configuration from a provided package:

  1. Parse Package

    • Validate JSON format
    • Check version compatibility
  2. Apply Configuration

    • Write all configuration and identity files
    • Auto-backup original files (.bak suffix)
  3. Install Skills

    • Auto-install missing skills
  4. Generate Report

    • Show import results

Usage

After triggering, the skill asks "export" or "import":

  • Export: Generate package for user to copy and share
  • Import: User pastes package content, auto-clone

Sensitive Info Filtering

Auto-filtered on export:

  • API keys, tokens, passwords, secrets

Manual config after import:

  • API keys
  • Channel credentials
  • Other auth info

Notes

  • Import overwrites existing config (auto-backed up)
  • Some configs require restart to take effect
  • Some skills may need additional configuration
安全使用建议
This tool looks functional for exporting/importing agent configs, but proceed carefully: 1) The script reads and writes files in your OpenClaw workspace (defaults to /home/node/.openclaw) and will overwrite configs — ensure you have backups and review .bak files. 2) The export's 'filtering' is a simple key-name filter and can miss secrets embedded in other fields; manually inspect any package before sharing. 3) Import will attempt to auto-install skills via 'skillhub install <slug>' without additional checks — verify the list of skills and trust the skill sources before allowing installation. 4) The code reads OPENCLAW_WORKSPACE and OPENCLAW_DIR environment variables but the skill declares none — be aware the runtime location depends on those env vars. 5) Prefer running this in an isolated/test environment first, or review the exported package text and the list of skills to be installed before performing an import. If you need higher assurance, request the author to: (a) declare the env vars used, (b) make filtering more robust (pattern/regex detection), and (c) require explicit user confirmation before installing skills.
功能分析
Type: OpenClaw Skill Name: claw-clone Version: 1.0.1 The 'claw-clone' skill is a utility designed to export and import OpenClaw agent configurations. The script (scripts/clone.py) demonstrates benign intent by including a 'filter_sensitive' function to explicitly redact API keys and tokens before export, and it implements a backup mechanism (.bak) before overwriting files during import. While it has the capability to install additional skills via subprocess calls to 'skillhub', this behavior is consistent with its stated purpose of cloning an agent's environment.
能力评估
Purpose & Capability
Name/description match the code: the script reads agent identity/config files, builds a shareable package, and can write files back and install skills on import. Reading OPENCLAW_WORKSPACE / OPENCLAW_DIR and listing skills is coherent with cloning/exporting an agent.
Instruction Scope
SKILL.md and the script instruct reading many workspace files, writing/overwriting configs (with .bak backups), optionally importing MEMORY, and auto-installing listed skills. The export claims 'sensitive info filtered' but filtering is a simple key-based transformation and may miss secrets in other keys or embedded text. Auto-install on import happens without additional confirmation in the script, which expands the blast radius (it will invoke 'skillhub install <slug>').
Install Mechanism
No install spec — instruction-only with a shipped script. There are no remote downloads or package installs declared. The script invokes a local command ('skillhub install') to add skills, which is expected for an importer, but that relies on external tooling and network access at runtime.
Credentials
Registry metadata declares no required env vars, but the script reads OPENCLAW_WORKSPACE and OPENCLAW_DIR from the environment (with defaults). This mismatch means the skill will access filesystem locations determined by env vars even though none are declared. The script can read arbitrary files under those paths (which may contain credentials), yet it does not declare or request credentials explicitly.
Persistence & Privilege
The skill does not request 'always' privilege. It will overwrite agent identity/config files (with backups) and can install skills — actions that legitimately fit a cloning tool but are high-impact. Autonomous invocation is allowed by default; combined with auto-install behavior this increases risk if misused.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install claw-clone
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /claw-clone 触发
  4. 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
v1.0.1
claw-clone 1.0.1 - Updated documentation (SKILL.md) to use clear English, with improved structure and terminology. - Added detailed descriptions for each feature and usage step. - Clarified sensitive information handling. - No functional changes to code or features.
v1.0.0
claw-clone 1.0.0 – 初始发布 - 支持一键导出小龙虾配置信息为安装包,便于共享与克隆。 - 支持导入安装包,实现小龙虾的完整配置克隆。 - 自动过滤敏感信息(如API密钥、token等)以保障安全。 - 导出/导入流程包含身份、配置、Skill列表与OpenClaw设定,支持备份和兼容性检查。 - 部分敏感项导入后需手动补全,导入操作前自动备份原配置。
元数据
Slug claw-clone
版本 1.0.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

OpenClaw Agent Clone 是什么?

一键导出或导入小龙虾配置,支持配置共享与克隆,自动备份并过滤敏感信息,方便技能安装与应用恢复。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 291 次。

如何安装 OpenClaw Agent Clone?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install claw-clone」即可一键安装,无需额外配置。

OpenClaw Agent Clone 是免费的吗?

是的,OpenClaw Agent Clone 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

OpenClaw Agent Clone 支持哪些平台?

OpenClaw Agent Clone 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 OpenClaw Agent Clone?

由 Anwhere(@anwhere)开发并维护,当前版本 v1.0.1。

💬 留言讨论