Open Source Stealth Browser MCP Fixes Agent Web Scraping
arhamislam5766 · hn · 2026-07-11
While using AI coding assistants, the author noticed that scraping web pages often consumes a massive amount of tokens (e.g., raw HTML from a single Wikipedia page hitting 68k tokens) and frequently fails on JS-rendered or anti-bot protected sites.
To solve this, the author developed an open-source stealth browser (based on a recompiled Chromium) and released it as an MCP for tools like Claude Code and Cursor.
- Results: The tool effectively bypasses certain anti-scraping checks and returns stripped-down content. For instance, previously failing JS test pages and the Nike website can now be successfully scraped with extremely low token consumption (around 285 and 700 tokens, respectively).
- Limitations: It currently lacks residential IP exits, meaning it can't breach advanced firewalls like Kasada. It's highly suitable for agent automation, QA testing, and data research.
More from coding & agent
- Soft Clamp cuts tool-call overuse in multi-teacher distillation, from 13.7% to 9.0% — antgroup · 2026-07-21
- SpecJudge runs locally on Ollama to pick the right-sized AI model for your project — jokiruiz · 2026-07-21
- A developer maps out six design rules for CLIs that humans and AI agents can both use — yujiezha · 2026-07-21
- A coding-agent skill that forces ADHD-friendly, answer-first output — ayghri · 2026-07-21
- A set of agent skills for CAD, robotics, and hardware design — earthtojake · 2026-07-21
- Outlines keeps LLMs on-rails with structured outputs — dottxt-ai · 2026-07-21