I’ve always found the “RLS enabled by default” behavior in Supabase/Postgres to be a double-edged sword: great for security, but brutal for DX when you’re just trying to get an agent or prototype working. The idea of having MCP servers enforce sane defaults automatically feels like a huge win for developer productivity without sacrificing safety.
Also really like that you’re tackling the “manual auth & secret wiring” problem — I’ve seen so many projects stall because the glue code around auth is brittle and error-prone.
Curious on two points:
How does InsForge handle multi-agent coordination? For example, if multiple AI agents are calling APIs concurrently, are there built-in isolation or quota mechanisms?
What’s your roadmap for integrations? (e.g., vector DBs, observability hooks, hosted dashboards)
This feels like it could lower the barrier for agentic apps quite a bit. Kudos on open-sourcing both client & server — looking forward to seeing how the community builds on this