OpenUser: Self-hosted user-persona tester for AI coding agents
OpenUser is a self-hosted, open-source tool that lets AI coding agents autonomously test features by simulating user personas in a browser, collecting console logs, network logs, and checkpoints. It runs locally with any coding agent and model, designed to replace manual testing in development loops.
Background
- The author is a developer building a large B2B ecommerce project and wants an AI agent that can do end-to-end browser testing for them, replacing manual testing after each code change.
- "Loop engineering" refers to the practice of setting up automated feedback loops where an AI coding agent (like GitHub Copilot or Cursor) writes code, tests it, fixes errors, and repeats — essentially a self-correcting development cycle.
- Existing tools like TestSprite offer AI-driven testing but cannot be self-hosted, meaning the user's code or data would go through a third-party cloud service.
- OpenUser is a new open-source, self-hosted CLI tool that creates simulated "user personas" who interact with the app in a browser, capturing console logs, network requests, and screenshots at each step. The AI agent can then use this feedback to autonomously fix issues.
- This fits a broader trend: developers trying to close the last mile of AI-assisted coding — moving from "AI writes code" to "AI also validates that code works correctly in a real browser environment."