When AI agents debug Supabase-backed apps, issues span the database, edge functions, auth, and client — making systematic investigation essential.
I built a structured debugging protocol: claude-debug
The approach: reproduce the exact failure, isolate with logging at each layer (client, edge function, RLS policy, database), confirm root cause, then fix.
https://github.com/krabat-l/claude-debug/blob/master/rules/debug.md
The user shares a structured debugging protocol for Supabase-backed applications, addressing issues across database, edge functions, auth, and client layers. They provide a GitHub link to their methodology, which involves reproducing failures, isolating issues with logging, confirming root causes, and fixing them.