Show HN: Sentinel – browser agent using 3x+ fewer tokens (open benchmark)

Category: library

Tags: ai-browser-agent, token-efficiency, benchmark, browser-automation, typescript

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

A benchmark comparing token efficiency of three AI browser automation frameworks. The project introduces a novel approach using Chrome's Accessibility Object Model to reduce token usage, achieving 3x-57x fewer tokens per task compared to alternatives, making it particularly interesting for cost-conscious AI agent development.

Target audience: backend devs

Repository: https://github.com/ArasHuseyin/browser-agent-benchmark · TypeScript · MIT

View on Hacker News