sd-advanced-search

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The content contains explicit anti-detection and CAPTCHA-bypass instructions (overriding navigator.webdriver and an automated Cloudflare Turnstile checkbox clicking workflow using cross-origin iframe snapshots), which are deliberate evasion techniques likely to facilitate abusive scraping or unauthorized access, even though there is no direct data exfiltration or backdoor code shown.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to navigate to and scrape live ScienceDirect search pages (using navigate_page, evaluate_script, DOM queries, and auto-clicking Cloudflare Turnstile), so it fetches and interprets untrusted public web content that can influence subsequent actions.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). This skill instructs the agent to actively bypass anti-bot protections (overriding navigator.webdriver and auto-clicking Cloudflare Turnstile), which is a direct attempt to circumvent security mechanisms on the browsing environment even though it doesn't request sudo or modify system files.