Changelog

Infrastructure Observability

Every InsForge project runs on a dedicated EC2 instance. You can now view real-time instance metrics — CPU utilization, memory usage, disk I/O, and network throughput — directly from the CLI with adjustable time ranges.

npx @insforge/cli diagnose metrics --range 1h

Backend Advisor

A new Backend Advisor automatically scans all active projects daily across security, performance, and health dimensions. Each scan produces actionable findings with severity levels (critical / warning / info) and concrete recommendations.

npx @insforge/cli diagnose advisor --severity critical

CLI Command

The new diagnose command brings all diagnostic capabilities together: instance metrics, advisor scans, database health checks (connections, slow queries, bloat, index usage, locks, cache hit ratio), and aggregated error logs from all backend sources.

npx @insforge/cli diagnose db --check slow-queries,connections
npx @insforge/cli diagnose logs

Agent Skill

We published an insforge-debug skill that teaches AI coding agents how to diagnose InsForge projects. It covers 9 common scenarios — SDK errors, HTTP status anomalies, edge function failures, slow queries, auth issues, realtime problems, performance degradation, and deploy failures — with proven diagnostic paths the agent follows automatically.

Read the full blog post

Edge Function Editor

You can now view, edit, download, and delete your edge functions directly from the dashboard. The built-in code editor gives you full visibility into your function source code without needing the CLI or an external editor.

Custom Domains for Project Websites

Developers can now add their own custom domains to project websites deployed with InsForge. Add a domain, follow the DNS records shown in the dashboard, and verify it directly in the deployment settings when propagation is complete.

Realtime Message Retention

InsForge Realtime now supports configurable message retention, so you can automatically clean up old realtime messages without manual maintenance. Choose a retention window like 30, 90, or 180 days, or keep messages forever, and InsForge will prune old messages on a daily schedule.

Deprecated: Hosted Auth Packages

The packages @insforge/react, @insforge/nextjs, and @insforge/react-router are now deprecated and should not be used in new projects. These packages provided pre-built authentication UI components that redirected users to backend-hosted auth pages using a cross-domain flow that passed tokens via URL query parameters.

What to use instead: Use @insforge/sdk directly for all authentication flows. Build your own auth UI and call SDK methods like signUp(), signInWithPassword(), signInWithOAuth(), and verifyEmail(). For SSR frameworks (Next.js, Remix, SvelteKit), use server mode with createClient({ isServerMode: true }) and manage tokens in httpOnly cookies.

View Auth SDK documentation

Open-Source Development Skills

We added open-source development skills to the InsForge platform repository. These skills guide AI agents contributing to the InsForge codebase with best practices for each package: backend, frontend, UI, shared-schemas, and docs. They follow the Agent Skills Open Standard and are available for Claude Code, Codex, and other compatible agents.

View InsForge repository

Bring Your Own Key (BYOK) for AI Gateway

We now support Bring Your Own Key (BYOK) for OpenRouter. You can configure your own API key directly in the AI gateway, giving you full control over usage, billing, and model access. The platform automatically handles key resolution across BYOK, managed, and environment-based credentials, so your applications continue to run without interruption.

We've also introduced a new Gateway Credentials UI to make key management simple and transparent, along with improved request handling that surfaces both response results and key sources for better debugging and visibility.