What problem does it solve? Keeping Lovable Preview tests synchronized with code changes is tedious: test plans drift from features, preview tokens get mishandled, and agents run full test suites when a small smoke run would suffice. This Skill standardizes how coding agents generate, run, and maintain Preview tests across Codex, Claude Code, and other Agent Skills runtimes. ## Core Features & Use Cases - Test Plan Management: Generate and sync test plans under .lovable-agent/tests/ with accurate covers: paths and last-sync commit tracking. - Scoped Test Runs: Execute the smallest affected plan set (--changed or --smoke) after implementation or deployment before offering --all. - Secure Token Handling: Store preview tokens only in the gitignored preview-token.local file, falling back to browser sessions or manual checklists when unavailable. - Workspace Migration: Migrate legacy .claude/lovable-claude/test/ workspaces with scripts/migrate-workspace.py while preserving read-compatible aliases. - Use Case: After deploying a new checkout flow, ask your agent to run the changed test plans against the live Preview app, marking any paid or destructive steps as [MANUAL]. ## Quick Start Ask your agent to run the changed Lovable Preview test plans for the features you just modified and report the results.
