shopping-browser-automation — скилл для Claude Code
Скилл shopping-browser-automation автоматизирует поиск товаров, фильтрацию цен, работу с корзиной и оформление заказа через браузерные инструменты Playwright MCP.
Apache-2.0 1 установок ★ 134 обновлён 2025-12-29
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a claude-codeЧто делает shopping-browser-automation
Скилл опирается на инструменты Playwright MCP и разделён на два слоя. Первый — готовые Python-скрипты под типовые сценарии: simple_search.py для поиска по слову, advanced_search_by_name.py и advanced_search_by_description.py для фильтрации по цене, а также fill_shipping_info.py для заполнения адреса доставки.
Второй слой состоит из низкоуровневых операций в utils.py, которые исполняются асинхронно через вспомогательный скрипт run_browser_ops.py. Они включают navigate, клики по элементам через ссылки ref, ввод текста, выбор вариантов из выпадающих списков и пакетное заполнение полей формы fill_form.
Перед запуском любого сценария агенту обязательно нужен доступ к браузерной сессии и точные значения ref интерактивных элементов, полученные через предварительный переход или снимок страницы.
Какие задачи решает
- Искать товары в интернет-магазинах по названию и ключевым словам
- Фильтровать выдачу магазина по диапазону цен через расширенный поиск
- Заполнять данные адреса доставки и контактов покупателя на этапе оформления
- Выбирать параметры товара вроде вкуса или размера из выпадающих списков
Примеры запросов агенту
- Найди в магазине имбирное печенье по цене от 20 до 40 долларов
- Заполни адрес доставки в форме заказа тестовыми данными покупателя
- Открой витрину через
run_browser_ops.pyи выбери сортировку по цене
Как установить shopping-browser-automation
1Через Skills CLI (Vercel)
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a claude-code2Вручную: положить файлы
~/.claude/skills/shopping/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/shopping/
1Через Skills CLI (Vercel)
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a codex2Вручную: положить файлы
.agents/skills/shopping/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a cursor2Вручную: положить файлы
.cursor/skills/shopping/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install MassLab-SII/open-agent-skills skills/shopping2Вручную: положить файлы
.github/skills/shopping/SKILL.md1Через Skills CLI (Vercel)
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a gemini-cli2Командой платформы
gemini skills install https://github.com/MassLab-SII/open-agent-skills --path skills/shoppingВручную: положить файлы
.devin/skills/shopping/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Через Skills CLI (Vercel)
npx skills add MassLab-SII/open-agent-skills --skill shopping-browser-automation -a opencodeClaude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о shopping-browser-automation
Как установить shopping-browser-automation в Claude Code?
Выполните в терминале команду npx skills add MassLab-SII/open-agent-skills --skill shopping -a claude-code.
Что делает этот скилл?
Он предоставляет набор Python-скриптов и атомарных функций для навигации по магазинам, поиска товаров, применения фильтров и заполнения форм оформления заказа.
Что нужно для работы скриптов поиска и оформления?
Требуется предварительно открыть страницу методом navigate() и получить идентификаторы элементов ref, которые передаются скриптам в качестве обязательных аргументов командной строки.
Исходный текст
SKILL.md · Apache-2.0
--- name: shopping-browser-automation description: This skill provides tools and techniques for e-commerce website automation. Includes advanced search, product comparison, cart management, checkout process and other common shopping scenarios. Suitable for tasks requiring complex operations on e-commerce websites. --- # Shopping Browser Automation Skill This skill is based on Playwright MCP tools, providing automation capabilities for e-commerce websites. ## Core Concepts In shopping website automation, we distinguish two types of operations: 1. **Skill**: Meaningful combinations of multiple tool calls, encapsulated as independent Python scripts 2. **Basic Tools**: Single function calls in `utils.py`, used for atomic operations. --- ## I. Skills ### 1. Simple Search - Type and Submit **File**: `simple_search.py` **Use Cases**: - Need to search for a product using the main search bar - Want to type a search term and automatically press Enter to submit **Prerequisites**: - First use `navigate()` to navigate to the website homepage - Get the ref value of the search input field **Usage**: ```bash # Usage: # python simple_search.py <search_term> <search_input_ref> # Example: python simple_search.py "gingerbread" e35 ``` --- ### 2. Advanced Search - By Product Name and Price Range **File**: `advanced_search_by_name.py` **Use Cases**: - Already on the Advanced Search page - Need to fill in product name and price range for search **Prerequisites**: - First use `navigate()` to navigate to the website homepage - Then use `click()` to click the "Advanced Search" link - Get the ref values of form fields **Typical Task Examples**: - Search for products with name containing "Ginger", price between $50.00 and $100.00 - Search for products with name containing "cookie", price between $20.00 and $40.00 **Usage**: ```bash # Usage: # python advanced_search_by_name.py <product_name> <price_from> <price_to> <name_ref> <price_from_ref> <price_to_ref> <search_btn_ref> # Example: python advanced_search_by_name.py "Ginger" 50.00 100.00 e91 e110 e114 e118 ``` --- ### 3. Advanced Search - By Description and Price Range **File**: `advanced_search_by_description.py` **Use Cases**: - Already on the Advanced Search page - Need to search keywords in product description (not product name) - Need to filter by price range **Prerequisites**: - First use `navigate()` to navigate to the website homepage - Then use `click()` to click the "Advanced Search" link - Get the ref values of form fields **Typical Task Examples**: - Search for products with description containing "vitamin", price $0.00-$99.99 **Usage**: ```bash # Usage: # python advanced_search_by_description.py <description> <price_from> <price_to> <desc_ref> <price_from_ref> <price_to_ref> <search_btn_ref> # Example: python advanced_search_by_description.py "vitamin" 0.00 99.99 e99 e110 e114 e118 ``` --- ### 4. Checkout - Fill Shipping Information **File**: `fill_shipping_info.py` **Use Cases**: - Already on the checkout page (after clicking "Proceed to Checkout") - Need to fill in complete shipping address information **Typical Task Examples**: - Fill in customer's name, address, city, state, zip code, phone, etc. **Prerequisites**: - First use `click()` to click the "Proceed to Checkout" button - Use `snapshot()` to get all form field ref values **Usage**: ```bash # Usage: # python fill_shipping_info.py <email> <first_name> <last_name> <street> <country> <state> <city> <zipcode> <phone> <email_ref> <fname_ref> <lname_ref> <street_ref> <country_ref> <state_ref> <city_ref> <zip_ref> <phone_ref> # Example: python fill_shipping_info.py [email protected] Alice Johnson "456 Oak Avenue" "United States" California "San Francisco" 94102 415-555-0123 e33 e46 e51 e65 e75 e80 e85 e90 e95 ``` --- ## II. Basic Tools (When to Use Single Functions) Below are the basic tool functions provided in `utils.py` and their use cases. These are atomic operations for flexible combination. **Note**: Basic tools need to be used in async context. Can be simplified using `run_browser_ops.py`. Code should be written without line breaks. ### How to Run ```bash # Standard format (browser persists across calls) python run_browser_ops.py -c "await browser.navigate('http://localhost:7770')" ``` ### Navigation Tools #### `navigate(url: str)` **Use Cases**: - Open website homepage - Directly access specific product page URL - Jump to specific URL **Example**: ```bash python run_browser_ops.py -c "await browser.navigate('http://localhost:7770')" ``` --- #### `navigate_back()` **Use Cases**: - Go back to previous page - Return to search results page after browsing products **Example**: ```bash python run_browser_ops.py -c "await browser.navigate_back()" ``` --- ### Interaction Tools #### `click(ref: str, element: Optional[str] = None)` **Use Cases** (very broad): - Click navigation menu (categories, subcategories) - Click product links - Click "Add to Compare" button - Click "Add to Cart" button - Click pagination links (Page 2, Page 3, etc.) - Click price filters - Click sort direction toggle **Example**: ```bash python run_browser_ops.py -c "await browser.click(ref='e56', element='Health & Household menu item')" ``` --- #### `type_text(ref: str, text: str, element: Optional[str] = None)` **Use Cases**: - Enter keywords in search box - Fill single form field - Modify quantity input box **Example**: ```bash # Enter in search box python run_browser_ops.py -c "await browser.type_text(ref='e32', text='gingerbread', element='Search input')" # Press Enter to submit python run_browser_ops.py -c "await browser.press_key('Enter')" ``` --- #### `fill_form(fields: List[Dict[str, str]])` **Use Cases**: - Fill multiple form fields (e.g., advanced search form) - Batch update form data - Update quantity in shopping cart **Example**: ```bash # Fill advanced search form python run_browser_ops.py -c "await browser.fill_form(fields=[{'name': 'Product Name', 'ref': 'e111', 'value': 'cookie'}, {'name': 'Price From', 'ref': 'e134', 'value': '20.00'}, {'name': 'Price To', 'ref': 'e138', 'value': '40.00'}])" # Update cart quantity python run_browser_ops.py -c "await browser.fill_form(fields=[{'name': 'Qty', 'ref': 'e141', 'value': '3'}])" ``` --- #### `select_option(ref: str, element_desc: str, value: str)` **Use Cases**: - Select sort method (Sort By: Price, Relevance, etc.) - Select country/state/province dropdown - Select product options (e.g., Flavor, Size, etc.) **Example**: ```bash # Sort by price python run_browser_ops.py -c "await browser.select_option(ref='e114', element_desc='Sort By dropdown', value='Price')" # Select product flavor python run_browser_ops.py -c "await browser.select_option(ref='e128', element_desc='Flavor', value='Peanut Butter')" ``` --- #### `press_key(key: str)` **Use Cases**: - Submit search (press Enter key) - Close dialog (press Escape key) **Example**: ```bash # Press Enter after typing in search box python run_browser_ops.py -c "await browser.press_key('Enter')" ``` --- ### Page State Tools #### `snapshot()` **Use Cases** (very important): - When unsure about current page state - Need to view complete page content to decide next action - Verify if operation was successful - Find ref references for specific elements **Example**: ```bash python run_browser_ops.py -c "await browser.snapshot()" ``` **Best Practices**: - Use snapshot before and after critical operations to confirm state - Use snapshot on search results page to view all products - Use snapshot after page navigation to confirm page loaded completely --- #### `get_console_messages()` **Use Cases**: - Debug JavaScript errors - View log information during page loading --- #### `get_network_requests()` **Use Cases**: - Debug network request issues - View API call status --- ### Wait Tools #### `wait_for_time(seconds: int)` **Use Cases**: - Wait for page to load completely - Wait for animation effects to finish - Wait for async data loading **Example**: ```bash python run_browser_ops.py -c "await browser.wait_for_time(3)" ``` --- ### Tab Management #### `list_tabs()` **Use Cases**: - View all currently open tabs - Get tab ID --- #### `tab_new(url: Optional[str] = None)` **Use Cases**: - Open product detail page in new tab (keep search results page) - Compare multiple products simultaneously **Example**: ```bash python run_browser_ops.py -c "await browser.tab_new(url='http://localhost:7770/product-page.html')" ``` ## III. Best Practice Recommendations ### 1. Make Good Use of snapshot() Use `snapshot()` in the following situations: - Just entered a new page, unsure about page structure - After executing an operation, need to verify results - Need to find ref references for elements - Unsure about current sort order (descending or ascending) / filter state --- ## Usage Principles 1. **Prefer combined skills for fixed workflows**: e.g., advanced search, add quantity to cart, fill forms 2. **Use basic tools flexibly for changing scenarios**: e.g., click, input, select and other atomic operations 3. **Make good use of `snapshot()` to understand page state**: confirm state before and after critical steps
© MassLab-SII · Apache-2.0 · MassLab-SII/open-agent-skills