lovable_skill

Overview

This skill integrates with Lovable.dev projects to generate exact deploy prompts and guidance for backend operations that GitHub sync cannot perform. It detects Supabase edge functions, migrations, and Lovable Cloud patterns and provides copy-paste prompts for deployment, migrations, RLS, storage, and auth changes. Use it when you need to trigger Lovable-managed actions or produce safe, actionable instructions.

How this skill works

The skill scans project structure for a supabase/ directory, Edge Functions, migration files, CLAUDE.md flags, and mentions of "Lovable" or deployment intents. When it detects backend changes that require Lovable Cloud action, it outputs an exact LOVABLE PROMPT ready to copy-paste into Lovable and optional warnings for destructive operations. It also recognizes yolo mode and indicates when automated browser-driven deployments are available versus manual prompt submission.

When to use it

• Working on a repo with a supabase/ directory or Edge Functions
• You add or edit supabase/migrations/*.sql and need to apply them
• You modify or add supabase/functions/*/index.ts and need to deploy the function
• You need to create tables, add columns, or change RLS policies in Lovable Cloud
• You need storage buckets, new secrets, or to enable OAuth providers

Best practices

• Only push backend changes to main so Lovable syncs them reliably
• After editing migrations or functions, provide the exact LOVABLE PROMPT to trigger deployment
• Include table schemas and column types explicitly when requesting new tables or migrations
• Use Yolo mode only when automated testing and browser automation are acceptable
• Keep secrets and env vars out of repo; add them via Lovable Cloud Secrets UI

Example use cases

• Deploy a single edge function: outputs: "Deploy the send-email edge function"
• Apply pending database migrations: outputs: "Apply pending Supabase migrations"
• Create a new table: outputs: "Create a profiles table with columns: id (uuid), user_id (uuid), name (text), created_at (timestamp)"
• Enable RLS for a table and allow authenticated users to read: outputs: "Enable RLS on profiles allowing authenticated users to read all rows"
• Create a private storage bucket: outputs: "Create a private storage bucket called user-uploads"