Some things I’m proud of in this release:
Skills are not config files. They are a composable entity that bundles connectors (SupaBase Table, Stripe, Posthog, any API or MCP), business logic, formatting rules, and preferences. Our agents read the skills as context, manually load them when selected based on user intent or on demand and even perform live analysis of your data schema when necessary to see if there are any misalignments or additional views, so the questions actually match the way your business works.
Onboarding scans your SupaBase schema and automatically generates a starter set of skills. From there, the Skill Builder is a simple language wizard that picks the right tables and connectors for what you’re trying to track.
Soon, skills will be accessible through CLI and MCP, so that other agents and platforms can consume them as well. If you’re building agentic tooling on top of SupaBase data, this is the part I’m most excited about.
Happy to learn about the architecture, the agent stack, or how we handle schema drift. Ask me anything.
<a href