August 2026 Summaries

Supabase has made enterprise-managed authentication for its MCP server generally available to Team and Enterprise customers using Okta SSO and Claude Team or Enterprise plans. Developed with Anthropic and Okta, the feature lets IT administrators centrally authorize and manage Supabase access in Claude through their identity provider, replacing individual OAuth approvals previously restricted to organization owners. Employees receive access based on their existing Supabase roles and project permissions, ensuring that queries and management actions in Claude cannot exceed their direct Supabase access. Organizations can restrict access through Okta groups, automatically remove Claude access during offboarding, and apply centralized security policies and reviews. The capability complements Supabase’s existing SSO, project-level roles, SOC 2, and HIPAA support, while SCIM-based platform provisioning is planned for the future.

Supabase-js now supports W3C Trace Context propagation, allowing client-side traces to share a trace_id with Supabase API Gateway and Edge Function logs for easier end-to-end debugging across browsers and backend services. Developers must configure their own OpenTelemetry tracer and provider, import the optional tracing entry point, enable tracePropagation when creating the Supabase client, and make requests within active spans. The integration sends standard traceparent, tracestate, and baggage headers only to Supabase and local development domains, while preserving bundle size and avoiding OpenTelemetry dependencies unless explicitly enabled. Trace IDs can also be included in Log Drains for correlation in external observability platforms such as Sentry, Datadog, Honeycomb, and Grafana. Limitations include no CDN or UMD support, support only for specified SDKs and log sources, and default respect for trace-sampling decisions. Users of earlier releases should upgrade to supabase-js 2.112.0 or newer because versions 2.106.0 through 2.111.x did not reliably send trace headers in bundled applications.

Supabase is available as a connector for Perplexity Computer, providing a persistent Postgres data layer that lets the agent read and write database records across runs without custom integration code. Perplexity Computer can perform multi-step tasks across more than 400 applications, including GitHub, Vercel, Datadog, Stripe, and Slack, enabling workflows that combine database queries with external tools. The connector can answer questions about production data, investigate support issues by reviewing user activity and payments, test Supabase Edge Functions, and schedule recurring reports for delivery to Slack. It can also combine Stripe cancellation data with Supabase behavioral data to identify customer drop-off patterns and create a GitHub pull request for a related fix. Users can create a free Supabase project if needed and connect it through Perplexity Computer.

Supabase has expanded Postgres Changes, its WebSocket-based system for receiving real-time insert, update, and delete events, with multi-column filtering, additional filter operators, and optional column selection. Filters can now combine multiple conditions with AND logic, enabling subscriptions such as tickets that are both open and assigned to a particular team, while new operators support patterns, null or boolean checks, regular expressions, NULL-safe inequality, and negation; a JavaScript filter builder is also available. Column selection allows subscribers to receive only needed fields, although primary keys are always included, reducing payload size and potentially lowering Realtime message and egress costs. Existing subscriptions remain unchanged unless these options are added, column selection requires supabase-js version 2.109.0 or later, and role permissions and Row Level Security continue to apply. OR logic, array and JSON containment, full-text search, and range operators are not currently supported, while DELETE events cannot use column filters because they only provide primary keys.

Supabase has open-sourced a tool called supabase/evals, designed to benchmark and assess the performance of AI agents, such as Claude Code and Codex, in handling real-world tasks within the Supabase ecosystem. This framework evaluates agents on tasks like building schemas and debugging Edge Functions, providing a scoring system that informs both a public benchmark and an internal regression suite. The initiative aims to measure the effectiveness of AI-driven development with Supabase, addressing the growing trend of using agents for project creation. By identifying areas where agents struggle, like inconsistent usage of declarative schemas and skill activation, Supabase seeks to refine its guidance and improve agent performance. The results are visualized on their website, offering insights into agent capabilities and areas needing improvement, with plans to enhance the tool further by expanding coverage and making scoring more stable.