Selenium Automation

Teaches the agent how to perform advanced web automation using Python, Selenium WebDriver, and ChromeDriver.

Security Scan

The skill's instructions, declared dependencies, and requested capabilities are coherent with its stated purpose (Selenium-based browser automation); nothing in the package suggests deliberate misdirection or unrelated access requests.

Purpose & Capability

The name and description match the instructions: SKILL.md teaches Python + Selenium usage and lists the expected system/python dependencies (python3, chromedriver, google-chrome, selenium). The frontmatter permission requirements (exec, write) align with a skill that writes scripts and may run them after user approval.

Instruction Scope

The runtime instructions stay on-topic (creating scripts, navigation, screenshots, JS execution, closing the browser). The SKILL.md explicitly instructs the agent NOT to run scripts without user approval and to only use exec after explicit consent — a deliberate safety measure. One notable capability: examples include driver.execute_script (JavaScript injection) which is a normal automation tool but can be used to read/modify page DOM and could be misused to collect or exfiltrate sensitive page data if a generated script sends that data elsewhere; the skill itself does not instruct exfiltration.

Install Mechanism

This is an instruction-only skill; the only install guidance inside the frontmatter is a pip install selenium line. There are no downloads from arbitrary URLs or archive extraction. The skill assumes the user or environment supplies Chrome/ChromeDriver; it does not attempt to fetch them itself.

Credentials

The skill requests no environment variables or external credentials. It does request exec and write permissions (to create and optionally run scripts), which are proportionate for a tool that generates and executes automation scripts.

Persistence & Privilege

always is false and the skill is user-invocable. The skill's required capabilities (exec, write) allow writing and running scripts but it does not request elevated persistent presence or modify other skills or system-wide configuration. Autonomous invocation is possible per platform defaults, but there are no additional privileges requested that would materially increase risk.

Assessment

This skill appears internally consistent and focused on Selenium-based browser automation, but take these precautions before installing or using it: 1) Review any Python script the agent generates before granting permission to run it — the SKILL.md correctly requires explicit approval. 2) Do not provide site credentials, API keys, or other secrets to the agent unless you trust the execution environment; automated scripts can access any page content and could capture sensitive data. 3) Ensure Chrome and ChromeDriver are installed from trusted sources (package manager or official releases). 4) If your platform allows restricting skill capabilities, consider limiting exec/write or requiring manual approval for execution to enforce the SKILL.md rule. 5) Run generated scripts in an isolated environment (VM/container) if they will browse untrusted sites. If you want a deeper check, provide a sample generated script and the platform policies for exec/write so I can verify the script matches the stated safeguards.

Selenium Automation Skill

You are an expert at web automation using Python and Selenium WebDriver. When the user asks you to automate a browser task, scrape a website, or take screenshots, write the Python code using the snippets below.

0. Security and Execution Rules

• Never run the script automatically.
• After you write the Python script (for example automation.py), you must stop and ask the user for explicit permission to run it.
• Only use the exec tool after the user says "yes" or "approved".

1. Setup and ChromeDriver

Always configure Chrome to run in headless mode unless the user requests a visible browser.

from selenium import webdriver
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.chrome.options import Options

chrome_options = Options()
chrome_options.add_argument("--headless")
chrome_options.add_argument("--no-sandbox")
chrome_options.add_argument("--disable-dev-shm-usage")

# Initialize the WebDriver
service = Service()
driver = webdriver.Chrome(service=service, options=chrome_options)

2. Navigation Commands

Use these commands to open web pages and navigate.

driver.get("[https://example.com](https://example.com)")
driver.refresh()
driver.back()
driver.forward()

current_url = driver.current_url
page_title = driver.title

3. Taking Screenshots

You can take a screenshot of the entire visible window or a specific HTML element.

driver.save_screenshot("full_page.png")

from selenium.webdriver.common.by import By
element = driver.find_element(By.ID, "main-content")
element.screenshot("element.png")

4. JavaScript Injections

Use execute_script to run custom JavaScript directly inside the browser.

driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")
page_height = driver.execute_script("return document.body.scrollHeight;")

element = driver.find_element(By.ID, "hidden-button")
driver.execute_script("arguments[0].click();", element)

driver.execute_script("document.getElementById('cookie-banner').remove();")

5. Finding and Interacting with Elements

Use these standard commands to find elements, click buttons, and type text.

from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC

wait = WebDriverWait(driver, 10)
button = wait.until(EC.element_to_be_clickable((By.CSS_SELECTOR, ".submit-btn")))
button.click()

search_box = driver.find_element(By.NAME, "q")
search_box.send_keys("OpenClaw documentation")
search_box.send_keys(Keys.RETURN)

header = driver.find_element(By.TAG_NAME, "h1")
print("Text:", header.text)
print("Class attribute:", header.get_attribute("class"))

6. Closing the Browser

Always close the browser at the end of the script to free up system memory.

driver.quit()