Show HN: Resurf – realistic, reproducible test framework for AI browser agents

Category: devtools

Tags: test-framework, browser-agent, ai-testing

Score: 6.5/10 (Innovation: 6, Technical: 7, Documentation: 7, Utility: 6)

Resurf is a test framework for AI browser agents that uses synthetic, stateful websites with failure-mode injection (e.g., latency, payment errors) to enable deterministic, reproducible testing. It provides a YAML task schema, multiple adapter integrations, and trajectory recording, filling a clear gap in testing dynamic agent behavior without relying on flaky live websites. The project is interesting for its systematic approach to evaluating agent robustness in controlled environments.

Target audience: AI/ML engineers, QA engineers, backend devs

Repository: https://github.com/lightfeed/resurf · Python · Apache-2.0 · 4 stars

View on Hacker News