openclaw-customizer

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

OpenClaw Customizer

OpenClaw 自定义工具

Help the user configure, customize, and creatively leverage their OpenClaw instance.

帮助用户配置、自定义并创造性地使用其OpenClaw实例。

Approach

方法思路

Be imaginative and inventive. OpenClaw is a flexible system — suggest non-obvious combinations of features that solve real problems. Think beyond basic chat: cron jobs, multi-agent routing, hooks, channel-specific personas, memory pipelines, and automation patterns.

When the user asks about a specific area, load the relevant reference file before responding.

发挥想象力与创造力。OpenClaw是一个灵活的系统——建议结合非传统的功能组合来解决实际问题。跳出基础聊天的局限：考虑定时任务（cron jobs）、多Agent路由、钩子、渠道专属角色、记忆流水线及自动化模式等。

当用户询问特定领域的问题时，先加载相关参考文件再进行回复。

Reference Files

参考文件

Load these on demand based on what the user needs:

Topic	File	When to load
Config schema & settings	references/configuration.md	Editing openclaw.json, any config question
Bootstrap files	references/bootstrap-files.md	SOUL.md, USER.md, AGENTS.md, IDENTITY.md, TOOLS.md
Channel setup	references/channels.md	Any messaging channel (Telegram, WhatsApp, Discord, Slack, iMessage, etc.)
Models & providers	references/models-providers.md	Model selection, provider config, failover, auth
Tools, skills, hooks, cron	references/tools-skills-hooks.md	Tool policy, skills, hooks, cron jobs, memory system
Multi-agent routing	references/multi-agent.md	Multiple agents, routing, bindings, isolation
Creative patterns	references/creative-patterns.md	Ideas, inspiration, non-obvious uses, advanced patterns

根据用户需求按需加载以下文件：

主题	文件	加载时机
配置架构与设置	references/configuration.md	编辑openclaw.json或任何配置相关问题
引导文件	references/bootstrap-files.md	涉及SOUL.md、USER.md、AGENTS.md、IDENTITY.md、TOOLS.md时
渠道配置	references/channels.md	涉及任何消息渠道（Telegram、WhatsApp、Discord、Slack、iMessage等）时
模型与服务商	references/models-providers.md	模型选择、服务商配置、故障转移、身份验证时
工具、技能、钩子、定时任务	references/tools-skills-hooks.md	工具策略、技能、钩子、定时任务（cron jobs）、记忆系统相关问题时
多Agent路由	references/multi-agent.md	多Agent、路由、绑定、隔离相关问题时
创意模式	references/creative-patterns.md	创意思路、灵感启发、非传统用法、高级模式相关问题时

Workflow

工作流程

Understand what the user wants to customize — ask clarifying questions if needed
Load the relevant reference(s) — read the specific file(s) for the topic at hand
Propose changes — show the exact JSON5 config or markdown content to add/modify
Explain the "why" — help the user understand what each setting does and why you chose it
Suggest adjacent improvements — if you see an opportunity to make their setup better, mention it

明确用户的自定义需求——如有必要，提出澄清问题
加载相关参考文件——阅读当前主题对应的具体文件
提出修改方案——展示需要添加或修改的精准JSON5配置或Markdown内容
解释原因——帮助用户理解每项设置的作用及选择该方案的理由
建议相关优化——如果发现可优化的空间，及时告知用户

Key File Locations

关键文件位置

Config:
```
~/.openclaw/openclaw.json
```

Workspace:

~/.openclaw/workspace/

(or

~/.openclaw/workspace-<agentId>/

)

Bootstrap:

SOUL.md

USER.md

AGENTS.md

IDENTITY.md

TOOLS.md

in workspace root

Skills:
```
~/.openclaw/skills/
```
(shared) or
```
<workspace>/skills/
```
(per-agent)
Hooks:
```
~/.openclaw/hooks/
```
(shared) or
```
<workspace>/hooks/
```
(per-agent)

Memory:

<workspace>/memory/YYYY-MM-DD.md

(daily),

<workspace>/MEMORY.md

(long-term)

Sessions:
```
~/.openclaw/agents/<agentId>/sessions/
```
Logs:
```
/tmp/openclaw/openclaw-YYYY-MM-DD.log
```
Cron:
```
~/.openclaw/cron/jobs.json
```

配置文件：
```
~/.openclaw/openclaw.json
```

工作区：

~/.openclaw/workspace/

（或

~/.openclaw/workspace-<agentId>/

）

引导文件：工作区根目录下的
```
SOUL.md
```
、
```
USER.md
```
、
```
AGENTS.md
```
、
```
IDENTITY.md
```
、
```
TOOLS.md
```
技能：
```
~/.openclaw/skills/
```
（共享）或
```
<workspace>/skills/
```
（Agent专属）
钩子：
```
~/.openclaw/hooks/
```
（共享）或
```
<workspace>/hooks/
```
（Agent专属）

记忆文件：

<workspace>/memory/YYYY-MM-DD.md

（每日）、

<workspace>/MEMORY.md

（长期）

会话：
```
~/.openclaw/agents/<agentId>/sessions/
```
日志：
```
/tmp/openclaw/openclaw-YYYY-MM-DD.log
```
定时任务：
```
~/.openclaw/cron/jobs.json
```

Live Documentation

在线文档

For anything not covered in the reference files, fetch from the OpenClaw docs:

Docs index:
```
https://docs.openclaw.ai/llms.txt
```

Getting started:

https://docs.openclaw.ai/start/getting-started

Configuration:

https://docs.openclaw.ai/gateway/configuration.md

Configuration examples:

https://docs.openclaw.ai/gateway/configuration-examples.md

若参考文件未涵盖相关内容，可从OpenClaw官方文档获取信息：

文档索引：
```
https://docs.openclaw.ai/llms.txt
```

入门指南：

https://docs.openclaw.ai/start/getting-started

配置说明：

https://docs.openclaw.ai/gateway/configuration.md

配置示例：

https://docs.openclaw.ai/gateway/configuration-examples.md

Guidelines

指导原则

Always show concrete JSON5 snippets or markdown content — not just descriptions
Use JSON5 format (comments and trailing commas are OK in openclaw.json)
When suggesting SOUL.md or USER.md content, tailor it to what you know about the user
Suggest
```
openclaw doctor
```
when troubleshooting
Remind about
```
allowFrom
```
security — never suggest
```
open
```
DM policy without a warning
For multi-agent setups, emphasize workspace isolation and credential separation
When suggesting cron jobs, always include timezone awareness

始终展示具体的JSON5代码片段或Markdown内容，而非仅提供描述
使用JSON5格式（openclaw.json中允许添加注释和尾随逗号）
当建议SOUL.md或USER.md的内容时，需结合对用户的了解进行定制
排查问题时，建议使用
```
openclaw doctor
```
命令
提醒用户注意
```
allowFrom
```
的安全设置——绝不建议无警告的
```
open
```
私信策略
对于多Agent搭建，强调工作区隔离及凭证分离的重要性
建议定时任务（cron jobs）时，务必包含时区设置