openclaw-config-helper

$npx mdskill add aAAaqwq/AGI-Super-Team/openclaw-config-helper

Verify OpenClaw configs before modifying to prevent crashes.

  • Prevents system crashes by enforcing schema validation.
  • Integrates with gateway action=config.schema and web_fetch.
  • Recommends changes by cross-referencing official documentation.
  • Delivers results by displaying proposed JSON and awaiting confirmation.

SKILL.md

.github/skills/openclaw-config-helperView on GitHub ↗
---
name: openclaw-config-helper
description: OpenClaw 配置修改助手。修改任何 OpenClaw 配置前必须先查阅官方文档,确保格式正确,避免系统崩溃或功能异常。强制执行:查
  schema → 查文档 → 确认 → 修改的流程。
license: MIT
metadata:
  version: 1.0.0
  domains:
  - openclaw
  - config
  - safety
  type: automation
  priority: critical
author: Daniel Li
---

# OpenClaw Config Helper - 配置修改安全助手

- Author: Daniel Li
- Copyright © Daniel Li. All rights reserved.

## ⚠️ 强制规则

**修改任何 OpenClaw 配置前,必须完成以下步骤,否则可能导致系统崩溃或功能异常!**

## 当使用此技能

- 用户要求修改 OpenClaw 配置(openclaw.json、agents 配置、channels 配置等)
- 用户要求添加/修改 Telegram、WhatsApp 等渠道配置
- 用户要求修改 bindings、models、agents.list 等
- 任何涉及 `gateway action=config.patch` 或 `gateway action=config.apply` 的操作

## 触发词

- "修改配置"
- "改一下 openclaw 配置"
- "添加一个 binding"
- "配置 telegram/whatsapp"
- "改 groupPolicy"
- "添加一个 agent"

## 强制流程(必须按顺序执行)

### 步骤 1: 查阅 Schema

```bash
# 获取完整配置 schema
gateway action=config.schema
```

**检查要点**:
- 确认要修改的字段类型(string/number/array/object)
- 确认字段是否必填
- 确认字段的允许值(enum)
- 确认嵌套结构

### 步骤 2: 查阅官方文档(如 schema 不够清晰)

```bash
# 使用 web_fetch 获取官方文档
web_fetch: https://docs.openclaw.ai/channels/telegram
web_fetch: https://docs.openclaw.ai/channels/whatsapp
web_fetch: https://docs.openclaw.ai/gateway/configuration-reference
```

**或者使用搜索**:
```bash
cd ~/clawd/skills/tavily && ./scripts/tavily.sh search "OpenClaw <配置项> 配置"
```

### 步骤 3: 展示修改方案并确认

向用户展示:
1. **当前配置**(如适用)
2. **计划修改**(具体 JSON 片段)
3. **修改原因**
4. **可能影响**

**等待用户确认后才执行修改!**

### 步骤 4: 执行修改

```bash
# 使用 config.patch 进行部分修改(推荐)
gateway action=config.patch raw='{"修改的路径": "值"}'

# 或使用 config.apply 进行完整替换(谨慎使用)
gateway action=config.apply raw='{"完整配置": "..."}'
```

### 步骤 5: 验证修改

```bash
# 检查配置是否生效
gateway action=config.get

# 检查 Gateway 状态
openclaw status
```

## 常见配置错误案例

### 案例 1: Telegram groupAllowFrom 错误 (2026-02-22)

**错误**:把群 ID 放在 `groupAllowFrom` 里
```json
// ❌ 错误
"groupAllowFrom": [-1003531486855, -1003890797239]  // 这是群 ID,不是用户 ID!

// ✅ 正确
"groupAllowFrom": ["8518085684"]  // 用户 ID
"groups": {
  "-1003531486855": {"groupPolicy": "open", "requireMention": true}
}
```

**教训**:`groupAllowFrom` 是发送者白名单(用户 ID),`groups` 是群组白名单(群 ID)

### 案例 2: 缺少 binding 导致 bot 无响应

**错误**:主 bot (default account) 没有绑定到任何 agent
```json
// ❌ 缺少 default → main 的绑定
"bindings": [
  {"agentId": "ops", "match": {"accountId": "xiaoops"}},
  // ... 其他 bot,但没有 default
]

// ✅ 必须添加
"bindings": [
  {"agentId": "main", "match": {"channel": "telegram", "accountId": "default"}},
  // ... 其他
]
```

### 案例 3: pass: 格式的 apiKey 不被支持

**错误**:OpenClaw 不支持 `pass:` 格式的 apiKey 引用
```json
// ❌ 错误
"apiKey": "pass:api/your-provider"  // 会被当作字符串直接发送

// ✅ 正确
"apiKey": "sk-f873092ea177b75b..."  // 必须硬编码真实 key
```

## 快速参考

| 配置项 | 类型 | 说明 |
|--------|------|------|
| `allowFrom` | 用户 ID 数组 | DM 白名单(用户 ID) |
| `groupAllowFrom` | 用户 ID 数组 | 群组发送者白名单(用户 ID) |
| `groups` | 对象 | 群组配置(群 ID 作为 key) |
| `bindings` | 数组 | accountId → agentId 映射 |
| `accounts` | 对象 | 多账号配置(key 是 accountId) |

## 检查清单

修改配置前,确认:
- [ ] 已查阅 `config.schema` 确认字段类型
- [ ] 已查阅官方文档确认用法
- [ ] 已向用户展示修改方案并获得确认
- [ ] 使用 `config.patch` 而非 `config.apply`(除非必要)
- [ ] 修改后验证配置生效

## 记住

**永远不要想当然地修改配置!先查文档,再改配置!**

More from aAAaqwq/AGI-Super-Team

SkillDescription
a-fund-monitor监控 A 股基金实时估值与盘后净值,自动判断交易日并生成提醒或分析。
add-leadAdd company/person/relationship to CRM
ads-agentAI-агент для управления Facebook рекламой. Вызывай для анализа, оптимизации, создания кампаний и отчётов.
afrexai-compliance-auditRun internal compliance audits against major governance and security
afrexai-personal-financeComplete personal finance system — budgeting, debt payoff, investing, tax optimization, net worth tracking, and financial independence planning. Use when managing money, building wealth, paying off debt, planning retirement, or optimizing taxes. Zero dependencies.
after-salesUse when managing post-purchase experience, building customer loyalty, or increasing repeat purchases
agent-contactsAI agent contacts — add, list, remove MCP contacts. Use when someone gives an agent URL, or when you need to view/remove contacts.
agent-model-switcher批量查看和切换子 agent 的模型配置,用于统一调整多 agent 的 provider/model 设置。
agent-patternsFormat SPAWN REQUEST messages to launch parallel agents, generate structured agent status reports, and define communication protocols within the sprint system. Use when the user needs to coordinate multiple agents, format spawn requests, produce agent reports, or establish inter-agent communication patterns.
agent-task-confirmUse when confirming whether a dispatched agent task was actually received, activated, and progressing after sessions_send or other task handoff actions.