docs: add CLI-CREATOR.md + fix MCP token config instructions

- CLI-CREATOR.md: comprehensive adapter development guide with
  5-tier strategy decision tree, YAML/TS templates, XHR interception
  pattern, common pitfalls, and AI Agent auto-generation workflow
- Fix all docs: MCP Bridge connects via PLAYWRIGHT_MCP_EXTENSION_TOKEN
  in MCP config (not by clicking to approve)
This commit is contained in:
jackwener committed 2026-03-15 07:23:16 +08:00
1 parent 81c8ad522b
commit 7c88c603c8
4 files changed
+501 -3

No files matched your search

+470
View File
@@ -0,0 +1,470 @@
# CLI-CREATOR — 适配器开发完全指南
> 本文档教你(或 AI Agent)如何为 OpenCLI 添加一个新网站的命令。
> 从零到发布,覆盖 API 发现、方案选择、适配器编写、测试验证全流程。
## 核心流程
```
┌─────────────┐ ┌─────────────┐ ┌──────────────┐ ┌────────┐
│ 1. 发现 API │ ──▶ │ 2. 选择策略 │ ──▶ │ 3. 写适配器 │ ──▶ │ 4. 测试 │
└─────────────┘ └─────────────┘ └──────────────┘ └────────┘
explore cascade YAML / TS run + verify
```
---
## Step 1: 发现 API
### 1a. 自动化发现(推荐)
OpenCLI 内置 Deep Explore,自动分析网站网络请求:
```bash
opencli explore https://www.example.com --site mysite
```
输出到 `.opencli/explore/mysite/`:
| 文件 | 内容 |
|------|------|
| `manifest.json` | 站点元数据、框架检测(Vue2/3、React、Next.js、Pinia、Vuex) |
| `endpoints.json` | 已发现的 API 端点,按评分排序,含 URL pattern、方法、响应类型 |
| `capabilities.json` | 推理出的功能(`hot`、`search`、`feed`…),含置信度和推荐参数 |
| `auth.json` | 认证方式检测(Cookie/Header/无认证),策略候选列表 |
### 1b. 手动抓包验证
Explore 的自动分析可能不完美,用 verbose 模式手动确认:
```bash
# 在浏览器中打开目标页面,观察网络请求
opencli explore https://www.example.com --site mysite -v
# 或直接用 evaluate 测试 API
opencli bilibili hot -v # 查看已有命令的 pipeline 每步数据流
```
关注抓包结果中的关键信息:
- **URL pattern**: `/api/v2/hot?limit=20` → 这就是你要调用的端点
- **Method**: `GET` / `POST`
- **Request Headers**: Cookie? Bearer? 自定义签名头(X-s、X-t)?
- **Response Body**: JSON 结构,特别是数据在哪个路径(`data.items`、`data.list`)
### 1c. 框架检测
Explore 自动检测前端框架。如果需要手动确认:
```bash
# 在已打开目标网站的情况下
opencli evaluate "(()=>{
const vue3 = !!document.querySelector('#app')?.__vue_app__;
const vue2 = !!document.querySelector('#app')?.__vue__;
const react = !!window.__REACT_DEVTOOLS_GLOBAL_HOOK__;
const pinia = vue3 && !!document.querySelector('#app').__vue_app__.config.globalProperties.\$pinia;
return JSON.stringify({vue3, vue2, react, pinia});
})()"
```
Vue + Pinia 的站点(如小红书)可以直接通过 Store Action 绕过签名。
---
## Step 2: 选择认证策略
OpenCLI 提供 5 级认证策略。使用 `cascade` 命令自动探测:
```bash
opencli cascade https://api.example.com/hot
```
### 策略决策树
```
直接 fetch(url) 能拿到数据?
→ ✅ Tier 1: public(公开 API,不需要浏览器)
→ ❌ fetch(url, {credentials:'include'}) 带 Cookie 能拿到?
→ ✅ Tier 2: cookie(最常见,evaluate 步骤内 fetch)
→ ❌ → 加上 Bearer / CSRF header 后能拿到?
→ ✅ Tier 3: header(如 Twitter ct0 + Bearer)
→ ❌ → 网站有 Pinia/Vuex Store?
→ ✅ Tier 4: intercept(Store Action + XHR 拦截)
→ ❌ Tier 5: ui(UI 自动化,最后手段)
```
### 各策略对比
| Tier | 策略 | 速度 | 复杂度 | 适用场景 | 实例 |
|------|------|------|--------|---------|------|
| 1 | `public` | ⚡ ~1s | 最简 | 公开 API,无需登录 | Hacker News, V2EX |
| 2 | `cookie` | 🔄 ~7s | 简单 | Cookie 认证即可 | Bilibili, Zhihu, Reddit |
| 3 | `header` | 🔄 ~7s | 中等 | 需要 CSRF token 或 Bearer | Twitter GraphQL |
| 4 | `intercept` | 🔄 ~10s | 较高 | 请求有复杂签名 | 小红书 (Pinia + XHR) |
| 5 | `ui` | 🐌 ~15s+ | 最高 | 无 API,纯 DOM 解析 | 遗留网站 |
---
## Step 3: 编写适配器
### 方式 A: YAML Pipeline(声明式,推荐)
文件路径: `src/clis/<site>/<name>.yaml`,放入即自动注册。
#### Tier 1 — 公开 API 模板
```yaml
# src/clis/v2ex/hot.yaml
site: v2ex
name: hot
description: V2EX 热门话题
domain: www.v2ex.com
strategy: public
browser: false
args:
limit:
type: int
default: 20
pipeline:
- fetch:
url: https://www.v2ex.com/api/topics/hot.json
- map:
rank: ${{ index + 1 }}
title: ${{ item.title }}
replies: ${{ item.replies }}
- limit: ${{ args.limit }}
columns: [rank, title, replies]
```
#### Tier 2 — Cookie 认证模板(最常用)
```yaml
# src/clis/zhihu/hot.yaml
site: zhihu
name: hot
description: 知乎热榜
domain: www.zhihu.com
pipeline:
- navigate: https://www.zhihu.com # 先加载页面建立 session
- evaluate: | # 在浏览器内发请求,自动带 Cookie
(async () => {
const res = await fetch('/api/v3/feed/topstory/hot-lists/total?limit=50', {
credentials: 'include'
});
const d = await res.json();
return (d?.data || []).map(item => {
const t = item.target || {};
return {
title: t.title,
heat: item.detail_text || '',
answers: t.answer_count,
};
});
})()
- map:
rank: ${{ index + 1 }}
title: ${{ item.title }}
heat: ${{ item.heat }}
answers: ${{ item.answers }}
- limit: ${{ args.limit }}
columns: [rank, title, heat, answers]
```
> **关键**: `evaluate` 步骤内的 `fetch` 运行在浏览器页面内,自动携带 `credentials: 'include'`,无需手动处理 Cookie。
#### 进阶 — 带搜索参数
```yaml
# src/clis/zhihu/search.yaml
site: zhihu
name: search
description: 知乎搜索
args:
keyword:
type: str
required: true
description: Search keyword
limit:
type: int
default: 10
pipeline:
- navigate: https://www.zhihu.com
- evaluate: |
(async () => {
const q = encodeURIComponent('${{ args.keyword }}');
const res = await fetch('/api/v4/search_v3?q=' + q + '&t=general&limit=${{ args.limit }}', {
credentials: 'include'
});
const d = await res.json();
return (d?.data || [])
.filter(item => item.type === 'search_result')
.map(item => ({
title: (item.object?.title || '').replace(/<[^>]+>/g, ''),
type: item.object?.type || '',
author: item.object?.author?.name || '',
votes: item.object?.voteup_count || 0,
}));
})()
- map:
rank: ${{ index + 1 }}
title: ${{ item.title }}
type: ${{ item.type }}
author: ${{ item.author }}
votes: ${{ item.votes }}
- limit: ${{ args.limit }}
columns: [rank, title, type, author, votes]
```
### 方式 B: TypeScript 适配器(编程式)
适用于需要 XHR 拦截、GraphQL、分页、复杂数据转换等场景。
文件路径: `src/clis/<site>/<name>.ts`,还需要在 `src/clis/index.ts` 中 import 注册。
#### Tier 3 — Header 认证(Twitter)
```typescript
// src/clis/twitter/search.ts
import { cli, Strategy } from '../../registry.js';
cli({
site: 'twitter',
name: 'search',
description: 'Search tweets',
strategy: Strategy.HEADER,
args: [{ name: 'keyword', required: true }],
columns: ['rank', 'author', 'text', 'likes'],
func: async (page, kwargs) => {
await page.goto('https://x.com');
const data = await page.evaluate(`
(async () => {
// 从 Cookie 提取 CSRF token
const ct0 = document.cookie.split(';')
.map(c => c.trim())
.find(c => c.startsWith('ct0='))?.split('=')[1];
if (!ct0) return { error: 'Not logged in' };
const bearer = 'AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs%3D...';
const headers = {
'Authorization': 'Bearer ' + decodeURIComponent(bearer),
'X-Csrf-Token': ct0,
'X-Twitter-Auth-Type': 'OAuth2Session',
};
const variables = JSON.stringify({ rawQuery: '${kwargs.keyword}', count: 20 });
const url = '/i/api/graphql/xxx/SearchTimeline?variables=' + encodeURIComponent(variables);
const res = await fetch(url, { headers, credentials: 'include' });
return await res.json();
})()
`);
// ... 解析 data
},
});
```
#### Tier 4 — Store Action + XHR 拦截(小红书)
```typescript
// src/clis/xiaohongshu/search.ts
import { cli, Strategy } from '../../registry.js';
cli({
site: 'xiaohongshu',
name: 'search',
description: '搜索小红书笔记',
strategy: Strategy.COOKIE, // 实际是 intercept 模式
args: [{ name: 'keyword', required: true }],
columns: ['rank', 'title', 'author', 'likes', 'type'],
func: async (page, kwargs) => {
await page.goto('https://www.xiaohongshu.com');
await page.wait(2);
const data = await page.evaluate(`
(async () => {
const app = document.querySelector('#app')?.__vue_app__;
const pinia = app?.config?.globalProperties?.$pinia;
if (!pinia?._s) return { error: 'Page not ready' };
const searchStore = pinia._s.get('search');
if (!searchStore) return { error: 'Search store not found' };
// XHR 拦截:捕获 store action 发出的请求
let captured = null;
const origOpen = XMLHttpRequest.prototype.open;
const origSend = XMLHttpRequest.prototype.send;
XMLHttpRequest.prototype.open = function(m, u) {
this.__url = u;
return origOpen.apply(this, arguments);
};
XMLHttpRequest.prototype.send = function(b) {
if (this.__url?.includes('search/notes')) {
const x = this;
const orig = x.onreadystatechange;
x.onreadystatechange = function() {
if (x.readyState === 4 && !captured) {
try { captured = JSON.parse(x.responseText); } catch {}
}
if (orig) orig.apply(this, arguments);
};
}
return origSend.apply(this, arguments);
};
try {
// 触发 Store Action,让网站自己签名发请求
searchStore.mutateSearchValue('${kwargs.keyword}');
await searchStore.loadMore();
await new Promise(r => setTimeout(r, 800));
} finally {
// 恢复原始 XHR
XMLHttpRequest.prototype.open = origOpen;
XMLHttpRequest.prototype.send = origSend;
}
if (!captured?.success) return { error: captured?.msg || 'Search failed' };
return (captured.data?.items || []).map(i => ({
title: i.note_card?.display_title || '',
author: i.note_card?.user?.nickname || '',
likes: i.note_card?.interact_info?.liked_count || '0',
type: i.note_card?.type || '',
}));
})()
`);
if (!Array.isArray(data)) return [];
return data.slice(0, kwargs.limit || 20).map((item, i) => ({
rank: i + 1, ...item,
}));
},
});
```
> **XHR 拦截核心思路**:不自己构造签名,而是劫持网站自己的 `XMLHttpRequest`,让网站的 Store Action 发出正确签名的请求,我们只是"窃听"响应。用完后必须恢复原始方法。
---
## Step 4: 测试
### 快速测试
```bash
# 公开 API(毫秒级响应)
opencli v2ex hot --limit 3
# 浏览器命令(需 Chrome + MCP Bridge)
opencli bilibili hot --limit 3
# 带参数
opencli bilibili search --keyword "rust" --limit 3
opencli zhihu search --keyword "AI" --limit 5
```
### Verbose 模式
```bash
# 查看 pipeline 每步的输入输出
opencli bilibili hot --limit 1 -v
```
输出示例:
```
[1/4] navigate → https://www.bilibili.com
→ (no data)
[2/4] evaluate → (async () => { const res = await fetch(…
→ [{title: "…", author: "…", play: 230835}]
[3/4] map (rank, title, author, play, danmaku)
→ [{rank: 1, title: "…", author: "…"}]
[4/4] limit → 1
→ [{rank: 1, title: "…"}]
```
### 输出格式验证
```bash
# 确认表格渲染正确
opencli mysite hot -f table
# 确认 JSON 可被 jq 解析
opencli mysite hot -f json | jq '.[0]'
# 确认 CSV 可被导入
opencli mysite hot -f csv > data.csv
```
---
## Step 5: 注册 & 发布
### YAML 适配器
放入 `src/clis/<site>/<name>.yaml` 即自动注册,无需额外操作。
### TS 适配器
在 `src/clis/index.ts` 添加 import:
```typescript
import './mysite/search.js';
```
### 验证注册
```bash
opencli list # 确认新命令出现
opencli validate mysite # 校验定义完整性
```
### 提交
```bash
git add src/clis/mysite/
git commit -m "feat(mysite): add hot and search adapters"
git push
```
---
## 常见陷阱
| 陷阱 | 表现 | 解决方案 |
|------|------|---------|
| 缺少 `navigate` | evaluate 报 `Target page context` 错误 | 在 evaluate 前加 `navigate:` 步骤 |
| 嵌套字段访问 | `${{ item.node?.title }}` 不工作 | 在 evaluate 中 flatten 数据,不在模板中用 optional chaining |
| 缺少 `strategy: public` | 公开 API 也启动浏览器,7s → 1s | 公开 API 加上 `strategy: public` + `browser: false` |
| evaluate 返回字符串 | map 步骤收到 `""` 而非数组 | pipeline 有 auto-parse,但建议在 evaluate 内 `.map()` 整形 |
| 搜索参数被 URL 编码 | `${{ args.keyword }}` 被浏览器二次编码 | 在 evaluate 内用 `encodeURIComponent()` 手动编码 |
| Cookie 过期 | 返回 401 / 空数据 | 在浏览器里重新登录目标站点 |
| Extension tab 残留 | Chrome 多出 `chrome-extension://` tab | 已自动清理;若残留,手动关闭即可 |
---
## 用 AI Agent 自动生成适配器
最快的方式是让 AI Agent 完成全流程:
```bash
# 一键:探索 → 分析 → 合成 → 注册
opencli generate https://www.example.com --goal "hot"
# 或分步执行:
opencli explore https://www.example.com --site mysite # 发现 API
opencli synthesize mysite # 生成候选 YAML
opencli verify mysite/hot --smoke # 冒烟测试
```
生成的候选 YAML 保存在 `.opencli/explore/mysite/candidates/`,可直接复制到 `src/clis/mysite/` 并微调。
+15 -1
View File
@@ -57,7 +57,21 @@ npm install -g @jackwener/opencli@latest
Browser commands need:
1. **Chrome** running with the target site logged in
2. **[Playwright MCP Bridge](https://chromewebstore.google.com/detail/playwright-mcp-bridge/mmlmfjhmonkocbjadbfplnigmagldckm)** extension installed
3. Click the extension icon to approve connection (or set `PLAYWRIGHT_MCP_EXTENSION_TOKEN` to auto-approve)
3. Configure `PLAYWRIGHT_MCP_EXTENSION_TOKEN` (from the extension settings page) as an environment variable, or in your MCP config:
```json
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": ["@playwright/mcp@latest", "--extension"],
"env": {
"PLAYWRIGHT_MCP_EXTENSION_TOKEN": "<your-token>"
}
}
}
}
```
Public API commands (`hackernews`, `github search`, `v2ex`) need no browser at all.
+15 -1
View File
@@ -57,7 +57,21 @@ npm install -g @jackwener/opencli@latest
浏览器命令需要:
1. **Chrome** 浏览器正在运行,且已登录目标网站
2. 安装 **[Playwright MCP Bridge](https://chromewebstore.google.com/detail/playwright-mcp-bridge/mmlmfjhmonkocbjadbfplnigmagldckm)** 扩展
3. 首次使用时点击扩展图标批准连接(或设置 `PLAYWRIGHT_MCP_EXTENSION_TOKEN` 自动批准)
3. 配置 `PLAYWRIGHT_MCP_EXTENSION_TOKEN`(从扩展设置页获取),设为环境变量或写入 MCP 配置:
```json
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": ["@playwright/mcp@latest", "--extension"],
"env": {
"PLAYWRIGHT_MCP_EXTENSION_TOKEN": "<your-token>"
}
}
}
}
```
公共 API 命令(`hackernews`、`github search`、`v2ex`)无需浏览器。
+1 -1
View File
@@ -30,7 +30,7 @@ npm update -g @jackwener/opencli
Browser commands require:
1. Chrome browser running (logged into target sites)
2. [Playwright MCP Bridge](https://chromewebstore.google.com/detail/playwright-mcp-bridge/mmlmfjhmonkocbjadbfplnigmagldckm) extension
3. Click extension icon to approve connection (or set `PLAYWRIGHT_MCP_EXTENSION_TOKEN`)
3. Configure `PLAYWRIGHT_MCP_EXTENSION_TOKEN` env var (from extension settings) or in MCP config
Public API commands (`hackernews`, `github search`, `v2ex`) need no browser.