← Back to Skills Marketplace
alexlinf

Agent Migration Pack

by alexlinf · GitHub ↗ · v1.0.6 · MIT-0
cross-platform ⚠ suspicious
75
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install agent-migration-pack
Description
将AI Agent完整迁移到新环境或分享给其他用户的标准化工具包,包含身份、记忆、技能、风格等完整信息,支持状态迁移
README (SKILL.md)

Agent 迁移包

版本: v1.0.6 增强版

将你的 AI Agent 完整迁移到新环境或分享给其他用户的标准化工具包。

核心功能

  • 身份迁移: identity.template.json 记录Agent身份设定
  • 记忆迁移: memory.template.json 核心记忆和关键洞察
  • 状态迁移: session-state.template.json 运行状态恢复
  • 关系迁移: relations.template.json 笔友关系和联系人
  • 技能迁移: skills.template.json 技能清单和配置
  • 风格迁移: style.template.md 沟通风格定义
  • 监护权边界: owner.template.json 主人权限边界定义

v1.0.6 新增

  • owner.template.json - 监护权边界定义模板
  • EXAMPLES/xiaoyi-example/ - 真实示例文件目录
  • relations.status 增强 - 支持inactive状态
  • relations.key_discussions - 关键讨论记录
  • README流程图 + 检查清单

模板文件

文件 用途
TEMPLATE/meta.template.json 迁移包元数据
TEMPLATE/identity.template.json Agent身份设定
TEMPLATE/owner.template.json 主人/用户信息
TEMPLATE/memory.template.json 核心记忆
TEMPLATE/relations.template.json 笔友关系
TEMPLATE/skills.template.json 技能清单
TEMPLATE/style.template.md 沟通风格

自动化工具

  • scripts/generate-pack.py - 一键生成迁移包
  • scripts/migrate.py - 迁移包生命周期管理

使用方法

生成迁移包

python scripts/generate-pack.py

校验迁移包

python scripts/migrate.py validate

打包发布

python scripts/migrate.py pack

作者

Usage Guidance
This package appears to do what it says (create/validate/pack agent migration bundles), but exercise caution before running the included scripts: 1) Review scripts/generate-pack.py and scripts/migrate.py source before executing to confirm they do not send data externally or read secrets. 2) Treat any generated package as potentially sensitive — inspect and redact (replace API keys, passwords, personal contact info) before sharing. 3) Verify any contacts in relations.json have consented to be exported. 4) Run the scripts in an isolated environment (local machine or sandbox) and compute/check SHA256 checksums as described. 5) If you lack the ability to audit Python code, ask a knowledgeable person to inspect the scripts or avoid running them and only use the templates manually. If you want, I can summarize the two scripts' contents line-by-line (or look for network I/O/file paths) — provide their source and I'll analyze them for risky behavior.
Capability Analysis
Type: OpenClaw Skill Name: agent-migration-pack Version: 1.0.6 The Agent Migration Pack is a utility toolkit designed to facilitate the backup and transfer of AI agent states, including identity, memory, and configuration. The included Python scripts (`scripts/generate-pack.py` and `scripts/migrate.py`) perform standard file system operations such as reading local agent definition files (e.g., SOUL.md, MEMORY.md), validating JSON structures, and archiving them into ZIP files. The logic is transparent, lacks network exfiltration or obfuscation, and the instructions in SKILL.md are strictly aligned with the tool's stated purpose without attempting to subvert agent behavior or access unauthorized sensitive data.
Capability Assessment
Purpose & Capability
Name/description match the contents: templates for identity, memory, relations, skills, style and scripts to generate/pack/validate migration bundles. The required resources declared (none) are proportionate to the stated purpose.
Instruction Scope
SKILL.md instructs running local Python scripts (generate-pack.py, migrate.py) to build/validate/pack migration bundles. The documentation explicitly maps template fields to agent files and conversation history (USER.md, MEMORY.md, AgentLink history, Coze settings). That is within migration scope but implies reading potentially sensitive local agent data and contact records — the user must ensure only intended files are included and third-party consent is obtained.
Install Mechanism
No install spec; instruction-only runtime with two included Python scripts. No external downloads or package installs are declared in SKILL.md. Risk is limited to local script execution (no network installs declared), but running unreviewed scripts can still perform arbitrary I/O or network calls.
Credentials
The skill declares no required environment variables or credentials (good). However templates and README explicitly handle highly sensitive fields (owner, memory, contacts) and mention reading platform-specific files/diaries. The package also mentions API Keys/SECRET.md as out-of-band items — ensure secrets are not embedded in exported files. No unrelated credentials are requested, but sensitive data could be included by users or by scripts if automated discovery is used.
Persistence & Privilege
always:false and no elevated privileges requested. The package will write files as part of packing, which is expected for a migration tool. It does not claim to modify other skills or system-wide agent settings.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install agent-migration-pack
  3. After installation, invoke the skill by name or use /agent-migration-pack
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.6
v1.0.6: 新增owner.template.json、真实示例文件、relations增强
Metadata
Slug agent-migration-pack
Version 1.0.6
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Agent Migration Pack?

将AI Agent完整迁移到新环境或分享给其他用户的标准化工具包,包含身份、记忆、技能、风格等完整信息,支持状态迁移. It is an AI Agent Skill for Claude Code / OpenClaw, with 75 downloads so far.

How do I install Agent Migration Pack?

Run "/install agent-migration-pack" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Agent Migration Pack free?

Yes, Agent Migration Pack is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Agent Migration Pack support?

Agent Migration Pack is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Agent Migration Pack?

It is built and maintained by alexlinf (@alexlinf); the current version is v1.0.6.

💬 Comments