Release v0.1.56: Error Tracking, Multi-Model AI, and Customer Dashboards

March 3, 2026

What changed

• Sentry Error Tracking: Runtime errors across client, server, and edge runtimes are now automatically captured and reported. Includes tunnel routing to bypass ad-blockers and source map support for readable stack traces.
• Multi-Model AI Provider Support: The model registry now wires all the way to agent execution. Users can override per-agent model defaults (Claude, GPT, Gemini, etc.) via a new "AI Models" tab on product pages.
• Customer Self-Service Portal: Generated enterprise products now ship with a complete /dashboard/account section for subscription management, invoice history, usage analytics, and GDPR-compliant data export and deletion.

Why it matters

These three features close critical gaps for production readiness: observability without documentation-only error tracking, flexibility in AI model selection without reimplementation, and customer autonomy without custom engineering.

Technical notes

• Sentry configuration is gracefully disabled in CI and when SENTRY_DSN is absent — no forced dependencies.
• All three Sentry runtimes (client, server, edge) are separately configured with appropriate sample rates and integrations.
• The customer portal is generated automatically as part of the product template.