Show HN: Mochi.js: bun-native high-fidelity browser automation library
5 hours ago
- #browser-fingerprinting
- #web-automation
- #stealth-browsing
- Every fingerprint surface (canvas, WebGL, etc.) is derived from a single profile and seed pair via a 48-rule DAG to ensure relational consistency and avoid mismatched fingerprints like Mac UA with Linux WebGL.
- mochi.js is a Bun-native browser automation framework using raw CDP, providing coherent fingerprints that survive probes like getParameter(0x9245), with JIT-friendly injection and real Chrome network behavior via session.fetch().
- It synthesizes human-like interactions (clicks, typing, scrolling) using biomechanical models such as Bezier paths and Fitts' law, parameterized by profiles for realism, and replaces traditional hand-stitched pipelines with one library.
- The framework includes automated challenge handling (e.g., Turnstile auto-clicking), maintains baselines from real devices for CI validation, and requires no Node, Python, or proprietary dependencies.