repository·main·Indexed 21 days ago
https://github.com/joeseesun/qiaomu-opencli-skillsA collection of Claude Code Skills for OpenCLI providing CLI interfaces for 79+ websites and desktop applications. It enables AI agents to automate tasks such as searching, posting, and downloading content from platforms like Bilibili, Twitter/X, and YouTube, as well as controlling Electron-based apps like Cursor, ChatGPT, and Notion. Includes specialized skills for adapter repair (qiaomu-opencli-autofix) and browser automation.
The following specialized skills are available to extend the capabilities of OpenCLI:
The qiaomu-smart-search skill acts as an intelligent router that directs queries to the best available opencli search sources based on the topic and scenario. It is designed for searching, querying, or researching information across websites, social media, technical documentation, news, shopping, travel, finance, or Chinese-language content.
Core Workflow:
opencli to discover real-time capabilities.grok, doubao, or gemini).搜索摘要
- 网站：<site1> | 查询词：<term1> | 次数：<n>
- 网站：<site2> | 查询词：<term2>；<term3> | 次数：<n>
- 已跳过：<site3>，原因：达到频率上限The shopping skills in Qiaomu OpenCLI allow you to search for products, compare prices, find deals, check reviews, and retrieve results from regional e-commerce platforms.
Available platforms include:
amazon: Global product search, price references, and English-language e-commerce.smzdm: Domestic (China) deals, discounts, shopping guides, and product discussions.coupang: South Korean e-commerce product search.douban: Supplementary reviews for books, movies, and music to aid consumer decision-making.When no specific website is requested, choose exactly one of the following AI sources based on the context:
grok: Best for real-time discussions, English-language internet discourse, Twitter/X context, and trending topics.doubao: Best for Chinese-language contexts, ByteDance/Douyin ecosystem, lifestyle content, and Chinese trending topics/Q&A.gemini: Best for global web content, English-language documentation, general information retrieval, and background summaries.Constraint: Once an AI source has been queried for a specific problem, do not re-query it with different keywords. If the answer is insufficient, supplement with specialized sources instead.
If automated exploration fails, use these high-priority heuristics to find APIs:
.json): Try adding .json to the URL (e.g., /r/all.json on Reddit) to access clean REST data.__INITIAL_STATE__): For SSR sites (e.g., Xiaohongshu, Bilibili), use page.evaluate('() => window.__INITIAL_STATE__') to grab the full data tree from the window object.evaluate to trigger hidden Network Fetches.OpenCLI uses a 5-level authentication system to access different types of data and services:
When writing an adapter in ~/.opencli/clis/, select the appropriate Strategy based on how you access data:
| Strategy | When to use | browser: setting |
|---|---|---|
Strategy.PUBLIC | Using a public API, no auth required | false |
Strategy.COOKIE | Needs existing login cookies/session | true |
Strategy.UI | Direct DOM interaction/scraping | true |
Note: Always prefer Strategy.PUBLIC (API) over Strategy.UI (DOM) if an API is available.
// Example Strategy.PUBLIC implementation
import { cli, Strategy } from '@jackwener/opencli/registry';
cli({
site: 'hn',
name: 'top',
strategy: Strategy.PUBLIC,
browser: false,
// ...
});For sites that use complex JavaScript logic, XHR signatures, or state management like Pinia (e.g., Xiaohongshu), use Strategy.INTERCEPT.
Instead of trying to replicate complex request signatures, you can use page.installInterceptor(pattern) to hijack the site's own XMLHttpRequest or fetch calls.
Workflow:
strategy: Strategy.INTERCEPT and browser: true.await page.installInterceptor('url_pattern') to start listening for specific requests.page.autoScroll() or calling a Pinia store action via page.evaluate()).await page.getInterceptedRequests().import { cli, Strategy } from '@jackwener/opencli/registry';
cli({
site: 'xiaohongshu',
name: 'notifications',
description: '小红书通知',
domain: 'www.xiaohongshu.com',
strategy: Strategy.INTERCEPT,
browser: true,
args: [
{ name: 'type', type: 'str', default: 'mentions' },
{ name: 'limit', type: 'int', default: 20 },
],
columns: ['rank', 'user', 'action', 'content', 'note', 'time'],
func: async (page, kwargs) => {
await page.goto('https://www.xiaohongshu.com/notification');
await page.wait(3);
// Install interceptor for specific URL pattern
await page.installInterceptor('/you/');
// Trigger API via Pinia store action
await page.evaluate(`(async () => {
const app = document.querySelector('#app')?.__vue_app__;
const pinia = app?.config?.globalProperties?.$pinia;
const store = pinia?._s?.get('notification');
if (store?.getNotification) {
await store.getNotification('${kwargs.type}');
}
})()`);
const requests = await page.getInterceptedRequests();
if (!requests?.length) return [];
let results: any[] = [];
for (const req of requests) {
const items = req.data?.data?.message_list || [];
results.push(...items);
}
return results.slice(0, kwargs.limit).map((item, i) => ({
rank: i + 1,
user: item.user_info?.nickname || '',
action: item.title || '',
content: item.comment_info?.content || '',
}));
},
});When using the OpenCLI skills, you can optimize your results by selecting the appropriate source based on the user's intent. Use the following routing logic to decide which skill to invoke:
arxiv.stackoverflow.hackernews or reddit.gemini or grok. If the information provided by those models is insufficient, supplement the results using the specialized technical sources listed above.Choose your Strategy based on how easily the API can be called via a fetch request in the browser console:
Result of fetch(url) | Strategy | Implementation Detail |
|---|---|---|
| Works directly | Strategy.PUBLIC | Use browser: false |
Works with {credentials: 'include'} | Strategy.COOKIE | Use func() pattern |
| Works with Bearer/CSRF headers | Strategy.HEADER | Use TS func() |
Fails via fetch but works on page | Strategy.INTERCEPT | Use installInterceptor |