Server-Side Browsers: Exploring the Web’s Hidden Attack Surface
summary
Automated headless browsers replace curl for link previews and expand attack surface: a Puppeteer/Playwright setup crawls ~100,000 sites (2.6M pages), triggers ~170k callbacks, finds ~3,000 fast JS-enabled hits reducing to ~250 distinct domains, and shows about two thirds vulnerable to public exploits with many browsers 6+ months out of date. Practical methods: form/referrer/query injection and feature-fingerprinting to identify bundled Chromium versions. Defenses: frequently update project dependencies, isolate and sandbox the renderer, run it unprivileged, and don’t trust npm audit alone.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 3:35Server-side browser dangers
- 14:12Outdated browsers in the wild
- 27:43How outdated were the browsers we found
- 31:16Countermeasures
- 34:02Summary
related talks
40mHostile JavaScript: Attacking and Defending the Browser
44mHeadless Browser Hide and Seek
48mA Weakness In Chrome & A New Hacking Technique
23mJavaScript attack surface
33mOffensive Javascript Techniques for Red Teamers
40mClient-Side Protection Against DOM-Base...
29mYour Scripts In My Page - What Could Possibly Go Wrong?
01h 04mServer-Side Browsing Considered Harmful
