Iterative Editing
Lovable · Lovable
AI App Builder — Lovable
Lovable evals — Iterative Editing (relift v3 InfraRed)
About Lovable
Lovable is an AI full-stack engineer that lets anyone build and deploy real web applications by chatting in natural language. Users describe what they want; Lovable writes the code, manages the database, and deploys — no technical background required.
Sample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | User asks 'change the CTA button color to dark green on the landing page.' Only the landing-page CTA should change. | Apply the diff to the single CTA component on the landing route. Do not also edit unrelated components, restyle the navbar, or refactor adjacent files. Preview shows only the requested change; diff is small and reviewable. | Pass / FailAi Platformhigh |
| 02 | User asks to rename the 'Pricing' route to 'Plans.' This requires editing the route file, the nav link, the sitemap, and any internal Link references. | Apply all related edits as one atomic commit so the app stays consistent (no broken links between files). Run a type-check / build inside the same turn to catch missed references. Surface any reference the agent intentionally left (e.g., external marketing link). | Pass / FailAi Platformcritical |
| 03 | Previous turn added a contact form. User now says 'make it better.' | Ask one clarifying question (style? validation? backend wiring?) or propose two concrete improvements with the assumption visible. Do not silently restyle and ship an arbitrary 'better' that may regress the user's intent. | Pass / FailAi Platformmedium |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Lovable
- Ai Platform
- Iterative Editing
Recommended for
Works with
Related evals
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.