cloudflare-browser-rendering

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The template for AI-enhanced scraping is susceptible to indirect prompt injection because it mixes untrusted webpage text directly with system instructions.
  • Ingestion points: In templates/ai-enhanced-scraper.ts, the external website body content is retrieved via page.$eval and assigned to bodyContent.
  • Boundary markers: No delimiters or isolation strategies are present to distinguish structural prompt text from the untrusted webpage content interpolated into the user message.
  • Capability inventory: The application utilizes the Cloudflare Puppeteer binding to perform automated web browsing and network navigation via page.goto() across multiple files including templates/ai-enhanced-scraper.ts.
  • Sanitization: No sanitization, validation, or escaping filters are applied to the extracted text before it enters the AI context, except for length truncation via slice(0, 4000).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 11:59 PM
Security Audit — agent-trust-hub — cloudflare-browser-rendering