Use saved versions as anchors for evidence: identify the requirement, explain the affected behaviour, and show how a reviewer can compare the relevant states. A version label alone does not explain whether a change is correct.
Understand the project record
Lovable describes a project as one app within a workspace, with a single project owner. Project content includes chat, preview, code, version history, and settings; edits are saved automatically to version history.
That establishes a place where changes are recorded, not a guarantee that every saved state is a useful release milestone. As editorial practice, select meaningful comparison points and document why they matter. Do not infer a rollback procedure or recovery guarantee from the existence of version history.
Write a change-evidence card
Create one card for each reviewable requirement:
- Requirement: the user problem and acceptance condition.
- Earlier state: the saved version reference and observed behaviour.
- Later state: the saved version reference containing the intended change.
- Affected path: the page, input, or sequence under review.
- Unchanged behaviour: nearby functions that must remain stable.
- Reproduction: synthetic inputs and steps another person can repeat.
- Verdict: accepted, rejected, or awaiting evidence, with a reviewer and date.
Use references that the team can actually locate. Do not invent version identifiers or treat an AI-generated summary as proof that the running app matches the explanation.
Hypothetical comparison for a class directory
Imagine a Bengaluru class directory where selecting “Saturday” previously removed the selected neighbourhood. The requirement is to preserve neighbourhood selection when changing the day.
Use fictional records for Indiranagar and Jayanagar, each containing weekday and Saturday classes. In the earlier state, record whether the neighbourhood resets. In the later state, select Indiranagar, switch to Saturday, and verify that only matching fictional classes remain.
Also check that clearing the day filter still preserves the neighbourhood. This second check distinguishes a narrow fix from a broader change to filtering rules.
The evidence card should say what happened for both inputs. “Filters improved” is insufficient because it neither identifies the original defect nor demonstrates preservation of adjacent behaviour.
Pause review when evidence cannot be repeated
Stop approval if the reviewer cannot locate the comparison states or reproduce the stated result. Next, ask the project owner to assemble a smaller evidence packet rather than authorising unrelated edits.
If access procurement is blocking that review, inspect the coding catalog for access format and saved terms before buying. Marketplace account/code/token offers are not proof of official subscription equivalence, and provider documentation does not establish marketplace offer terms.
Sources and scope
Source material checked 29 October 2026; features/policies may change. Lovable project overview establishes project structure, ownership, and automatic version-history saving. The evidence card and fictional directory comparison are editorial review methods, not tested rollback instructions.