Skip to main content

​
v1.12.1 — March 12, 2026

​
Major runtime performance gains, stronger thread and chat recovery, and Snowflake compatibility fixes

​
Performance and Runtime Reliability

The runtime was reworked to reduce event-loop stalls, trim unnecessary thread contention, and take more Snowflake and metadata work off customer-facing request paths.
  • The server now uses an async-first completion path that sharply reduces loop lag and cuts the number of long-lived worker threads, improving responsiveness under concurrent chat load.
  • Snowflake hot-path work was reduced with queued message-log writes, batched lookups, runtime caches, and fewer synchronous metadata reads during thread creation and active chat flows.
  • Thread search and thread metadata handling were rebuilt to avoid expensive startup and request-path churn, which improves search consistency and reduces latency on busy systems.
  • Retry handling now summarizes oversized contexts before replay and cancels superseded retry backoff more cleanly, which helps prevent large retries and follow-up prompts from getting stuck behind stale work.

​
Snowflake and Environment Compatibility

Snowflake-hosted environments gained broader compatibility and safer startup behavior across app, workspace, and service flows.
  • Genesis now better handles Snowflake-hosted app environments, including PrivateLink SPCS hostnames, so UI, API, and WebSocket traffic stay aligned across more Snowflake deployment topologies.
  • Managed workspace backports were restored on the release line so workspace and artifact services can start cleanly again in environments that depend on managed app flows.
  • A build regression caused by stale Genesis Apps workspace imports was removed so affected environments can package and start cleanly again.

​
Chat, Threads, and UI Reliability

Several customer-facing issues around thread continuity, dashboard clarity, and refresh behavior were addressed across chat and monitoring surfaces.
  • New draft threads no longer confuse provisional client IDs with persisted backend threads, and deleted or stale plist logs can now self-heal instead of breaking thread or blueprint recovery.
  • Azure OpenAI relayed chats now recover more cleanly from streaming hangs, and completion waits are matched to the correct message so replies do not end early or come back empty.
  • Mission status changes and follow-up questions now stay synchronized more reliably during active work.
  • Live slot views and cost dashboards now surface user display names directly, fixing cases where operators saw incomplete identity information in active system views.
  • The Cost Usage dashboard refresh now rereads data in place, shows a clear busy state during manual refreshes, and avoids the flickering refresh indicator seen during routine polling.
  • App and dashboard tiles were cleaned up to handle dense metadata and tags more gracefully, and version display in upgrade messaging was simplified.
  • Chat and coding-agent restore flows no longer do unnecessary global scans on first turn, which reduces avoidable delays for threads that never used those features.

​
Genesis Apps and Platform Hardening

Genesis Apps, sandbox, and managed-runtime flows were hardened to close edge cases that previously leaked through to users or operators.
  • Genesis Apps import and runtime validation is stricter, blocking forbidden server endpoints earlier and hardening sandbox, auth, and config edge cases that could lead to failed starts or confusing recovery paths.
  • Uploads and generated image assets can now move into Genesis Apps workspaces more reliably, improving app-building workflows.
  • Genesis Apps dashboard scaffolding now has safer defaults and clearer dataset-first guidance for new app builds.
  • Mission and project metadata writes now flow through a local cache and durable outbox before Snowflake persistence, reducing customer-visible delays during active mission work.
  • Image, file, git, and managed-repo flows picked up follow-up fixes around path handling, persistence, and workspace routing that improve stability in more complex app-building scenarios.

​
Operational Visibility and Diagnostics

Operators now have better visibility into runtime health and the sources of performance degradation when a deployment comes under load.
  • Snowflake query observability and telemetry coverage were expanded, making it easier to identify which request paths are still consuming disproportionate database time.
  • Additional event-waiter, notifier, and server telemetry fixes improve runtime diagnostics for managed workspaces and live chat delivery behavior.

​
Upgrade Notes

Upgrade from v1.10.5 to v1.12.1 includes major runtime concurrency changes and Snowflake host-detection updates. Validate Snowflake-host routing and environment-specific startup behavior in lower environments before production rollout.