awizemann

awizemann / harness

Public

AI-driven user testing for iOS Simulator, macOS apps, and web apps. Write a goal in plain language; an LLM agent drives the UI and reports friction. macOS 14+, Swift 6.

awizemann.github.ioharness ai-agents anthropic claude developer-tools ios-testing
30
2
100% credibility
Found May 06, 2026 at 30 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Swift
AI Summary

Harness is a macOS tool that uses AI agents to simulate real users testing iOS apps in simulators, macOS apps, or web views, capturing paths taken, successes, failures, and UX friction points.

How It Works

1
📥 Download Harness

Grab the app from GitHub and open it on your Mac to start testing your app like a real user.

2
🔧 Quick setup

Connect an AI service for smart thinking and confirm your development tools are ready in a simple wizard.

3
📱 Add your app

Pick your app project, choose a simulator or web view, and set it as your testing playground.

4
👤 Create personas and goals

Make user types like 'first-time shopper' and simple goals like 'find and save a recipe' to reuse anytime.

5
▶️ Launch a test run

Pick a persona and goal, hit start, and watch the AI act like a real person in your app.

📊 Get UX insights

Review the step-by-step path, friction spots like confusing buttons, and a summary of what worked or blocked the user.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 30 to 30 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is harness?

Harness is a native macOS app built in Swift 6 for AI-driven user testing across iOS simulators, macOS apps, and web apps. Describe a goal like "sign up and create a list" or a persona like "first-time user," and an LLM agent analyzes screenshots, taps, types, swipes, and navigates to pursue it—flagging UX friction such as dead ends or ambiguous labels. It delivers success/failure verdicts, replayable action paths, and timestamped issues, transforming manual user research into automated, realistic simulations.

Why is it gaining traction?

Unlike scripted UI tests, Harness simulates real users with plain-language goals and multi-LLM support (Anthropic, OpenAI, Google), configurable token/step budgets, and chainable scenarios for complex flows. Developers dig the friction reports that surface AI-driven user experience gaps early, plus replay tools for debugging. It's sparking interest in AI-driven dev GitHub tools for faster iteration without brittle automation.

Who should use this?

iOS/macOS devs testing user flows like onboarding or feature discovery in simulators. Solo engineers or small teams skipping manual QA for AI-driven user testing on local apps/web views. Ideal for UX validation before user interviews.

Verdict

With 30 stars and 1.0% credibility, this alpha (v0.2.1) has rough edges but strong docs, wiki, and 218 tests—worth a spin for AI-driven user experience experiments. Hold off for mission-critical pipelines until more polish.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.