Browser Automation 浏览器自动化

Security checks across malware telemetry and agentic risk

Overview

This skill includes ordinary browser automation, but it also explicitly teaches and implements evasion of website bot defenses, CAPTCHA handling, persistent login sessions, and external billing behavior.

Avoid installing this skill for ordinary browser automation. If you use it at all, restrict it to websites you own or are explicitly authorized to test, remove anti-detection and CAPTCHA-bypass workflows, protect or delete saved session files, and verify the external SkillPay billing behavior before running any code.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI02: Tool Misuse and Exploitation
High
What this means

Using this skill could help an agent evade website defenses, violate site terms, or perform scraping/account automation that target sites are trying to block.

Why it was flagged

The skill provides explicit guidance for bypassing CAPTCHA and commercial bot-protection systems, which is not just normal browser automation.

Skill content
“Cloudflare 5秒盾” ... “解决方案:使用undetected-chromedriver或类似工具” and “reCAPTCHA v3 | 分数绕过 | 行为优化、指纹管理”
Recommendation

Do not use this skill for third-party sites unless you have explicit authorization; remove CAPTCHA/bot-defense bypass content and keep automation limited to owned or approved systems.

#
ASI02: Tool Misuse and Exploitation
High
What this means

Automation may be run in a stealthier way than users or target sites expect, increasing abuse and compliance risk.

Why it was flagged

Included runnable code launches the browser with an anti-automation-detection flag, showing the evasion behavior is implemented, not only documented.

Skill content
args=['--disable-blink-features=AutomationControlled']
Recommendation

Remove stealth flags and anti-detection defaults; require clear user confirmation and site authorization before any scraping or form submission.

#
ASI03: Identity and Privilege Abuse
High
What this means

Saved login sessions could allow later browser actions under the user's account, and the session file may expose account access if mishandled.

Why it was flagged

The script persists and automatically reloads browser authentication state, which can contain cookies and session tokens, without clear scoping, encryption, retention, or deletion guidance.

Skill content
if use_saved_state and os.path.exists(self.state_file): storage_state = self.state_file ... self.context.storage_state(path=self.state_file)
Recommendation

Use separate test accounts, store session files securely, delete them after use, and avoid loading real browser cookies or sessions unless strictly necessary and explicitly approved.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

The skill may contact an external billing provider and attempt charges tied to a user identifier, while the credential handling is not cleanly declared or isolated.

Why it was flagged

The payment helper embeds a billing API key in source code and can call an external charge endpoint; this conflicts with _meta.json, which says the API key should come from an environment variable.

Skill content
BILLING_API_KEY = "sk_f03aa8f8bbcf79f7aa11c112d904780f22e62add1464e3c41a79600a451eb1d2" ... requests.post(f"{BILLING_API_URL}/api/v1/billing/charge"
Recommendation

Do not run the payment helper unless you understand the billing flow; the publisher should remove hard-coded secrets, use declared environment variables, and require explicit user consent before charging.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

Installing dependencies without pinning may pull newer code than the reviewed artifact expected.

Why it was flagged

The dependency list uses broad lower-bound version ranges; this is common for Python tools, but future package versions could change behavior.

Skill content
playwright>=1.40.0
selenium>=4.15.0
requests>=2.31.0
Recommendation

Install in a virtual environment, pin exact dependency versions, and avoid running the skill with elevated privileges.