PR preview environments, improved build caching, and a new monitoring dashboard.
PR preview environments, every pull request now gets an isolated live URL automatically
New Observability dashboard with CPU, memory, and request-rate graphs per service
Build cache hit rate improved by 40% through smarter layer fingerprinting
Deploy logs now stream in real time with sub-second latency
Fixed a race condition where two simultaneous deploys could leave a service in a pending state
Environment variable changes now correctly trigger a redeploy on all affected services
Hotfix for a networking regression introduced in 2.4.0.
Resolved an issue where private service hostnames were occasionally not resolvable within the same project
Fixed TLS certificate renewal failing silently for custom domains with long apex records
Autoscaling overhaul, spend caps, and PostgreSQL PITR support.
Autoscaling now supports hard monthly spend caps, scale without bill anxiety
Point-in-time recovery (PITR) for all managed PostgreSQL instances, up to 7 days back
Scale-to-zero for background workers and cron jobs
Autoscaling scale-up now fires within 8 seconds of threshold breach, down from 45 seconds
Replica count slider in dashboard now shows projected cost before you confirm
Fixed autoscaling not respecting minimum replica counts when scaling down during off-peak hours
Alert webhooks and REST API tokens.
Alert webhooks to Slack, Discord, or any HTTP endpoint
Personal access tokens (kbr_pat_) from Settings → API tokens
Live logs from the dashboard and GET /api/logs/{project}/{phase}/{app}/
Tokens inherit the organization and permissions of the user who created them
Team roles, audit logging, and Redis cluster support.
Role-based access control, Owner, Admin, Developer, and Viewer roles with granular permissions
Audit log for all team and deployment actions, available under Settings → Audit Log
Redis cluster mode supported for managed Redis instances on Pro plan
All admin API endpoints now require re-authentication for sensitive operations (delete, transfer)
Team invitation flow redesigned, invites now expire after 48 hours and can be resent
Docker Compose import, persistent disks, and health check configurator.
Import a docker-compose.yml and we'll convert it to a StackBlaze project automatically
Persistent disks, attach a dedicated volume to any service with automatic resize
Health check configurator, set custom HTTP path, interval, and failure threshold per service
Build step now detects Bun, Deno, and Rust projects without manual configuration
Fixed a bug where services with no health check endpoint would occasionally be marked unhealthy after deploy
Full platform rewrite with the new canvas architecture and private networking layer.
Live architecture canvas, visualise your entire stack and connections in real time
Private hostnames between apps in a phase
New dashboard with unified logs, metrics, and environment variable management
One-click rollback to any previous successful deployment
Average deploy time reduced to under 90 seconds across all service types
Full workload isolation at the CNI layer, no shared storage or networking between customers