A county open-data GraphQL endpoint in production must not allow anonymous clients to explore the full schema. What should operators enforce?

Requiring authentication and disabling anonymous introspection on the production GraphQL endpoint is the hardening step county open-data operators should enforce. Even when datasets are intentionally public, anonymous introspection lets any internet client walk the full schema—types, fields, relationships, and often clues about internal operations—without proving identity. Binding the production endpoint to authenticated callers and turning off anonymous introspection shrinks the reconnaissance and abuse surface while still allowing approved open-data access patterns through controlled queries. Leaving introspection open to anonymous internet clients mistakes developer discoverability for a reason to expose production schema details. Publishing the GraphQL admin password in the open-data portal footer is credential disclosure, not a security control. Allowing open-data access to bypass role checks on every base-table mutation confuses public read access with write authority and risks integrity damage to municipal systems. Exam caveat: “open data” does not automatically mean open schema mutations or anonymous production introspection; public read projections and administrative GraphQL capabilities should remain distinct. Introspection may be acceptable in non-production with authentication, while production should deny anonymous schema exploration. Practical check: enforce authentication at the production GraphQL gateway, verify that an unauthenticated introspection request fails, confirm public queries return only approved fields, and check that mutations require elevated roles. Review gateway logs for introspection probes and unexpected mutation attempts.