Quality & Evaluation
App Evaluation
App Evaluation is Webase's built-in quality gate. It uses AI to review your app against a set of requirements, scores it from 0 to 100, and identifies specific issues that need fixing.
Free to use. Evaluations do not consume any of your monthly messages. You can run them as often as you like at no cost.
How Evaluation Works
When you run an evaluation, a separate AI reviewer examines your app's source files and checks them against a list of quality requirements. The reviewer looks at code structure, functionality, user experience, and adherence to best practices.
The evaluation produces three things:
- A score from 0 to 100 — An overall quality rating for your app.
- Per-requirement results — A breakdown showing whether each specific requirement passed or failed.
- Runtime issue detection — Identification of potential errors, broken imports, or logic problems that could cause your app to crash.
Running an Evaluation
Click Evaluate
In the Application Editor toolbar, click the Evaluate button. The evaluation begins immediately.
Wait for Results
The AI reviewer analyzes your app's files. This typically takes 10 to 30 seconds depending on the size of your app. You can continue working while the evaluation runs.
Review the Results
Once complete, you see the score and a detailed breakdown. Each requirement is listed with a pass or fail status and an explanation of what was found.
Understanding the Score
The evaluation score gives you a quick sense of your app's overall quality:
- 70 and above — Your app passes the quality gate. It meets the core requirements and should work well for users.
- 50 to 69 — Your app has some issues that should be addressed. Review the failed requirements for specific guidance.
- Below 50 — There are significant issues. The detailed breakdown will highlight what needs the most attention.
Pass threshold: 70. An evaluation result of 70 or higher is considered "passed." Below 70 is "failed." This threshold ensures a consistent quality baseline across all Webase apps.
What Gets Checked
The evaluation covers several categories of requirements:
- Functionality — Does the app do what it is supposed to do? Are all described features present and working?
- Code quality — Is the code well-structured? Are components organized logically?
- User experience — Is the interface intuitive? Are there clear navigation patterns and helpful feedback for user actions?
- Runtime safety — Are there broken imports, missing dependencies, or logic errors that could crash the app?
- Best practices — Does the app follow Webase conventions like using HashRouter, esm.sh imports, and Tailwind via CDN?
Acting on Evaluation Results
After reviewing your evaluation, you have several options:
- Fix issues manually — Use the chat to ask the AI to address specific failed requirements. For example: "Fix the broken import on the Dashboard component."
- Use Evaluate & Improve — Let the AI automatically fix issues and re-evaluate in a loop. See Evaluate & Improve for details.
- Ignore specific findings — If a requirement does not apply to your use case, you can note it and move on.
Evaluation History
Every evaluation is saved so you can track your app's quality over time. View past evaluations from the app settings to see how your score has changed and which issues have been resolved.