Skip to main content

v1.3.0

What's Changed

☸️ Kubernetes

  • feat: Sort Kubernetes clusters by connection health on the landing page
  • feat: Display live CPU and memory usage in pod container resource views
  • feat: Support cross-filtering dependencies between namespace and workload dropdowns across telemetry views
  • feat: Automatically run server-side recommendation scans on initial Kubernetes agent connection
  • fix: List GKE clusters across locations to prevent multi-region InvalidArgument errors

🔍 Troubleshoot

  • feat: Support GCP Cloud Trace evidence and trace-correlated log viewing
  • feat: Add Cloud Run deployment-diff card and structured cloud logs viewer
  • feat: Add series-match metric discovery capabilities for workload telemetry
  • feat: Route cost-server metrics via the observability API
  • feat: Support provider-independent canonical log fetching with end-to-end query, provider, and label metadata
  • feat: Expand Knowledge Graph with in-cluster database classification, Azure resource edge derivation, and region/zone node metadata
  • feat: Refactor alarm actions into an interactive side panel with configuration editing
  • feat: Support per-instance integration account mapping and display integration profiles in expandable user list rows
  • feat: Add LLM-assisted per-class triage scoring with tenant feature flags, status history timelines, and change reason tracking
  • feat: General UI enhancements including sticky table headers, brand-neutral partner loaders, 48-hour empty states, and tenant boundary banners
  • fix: Automatically rebuild shared RabbitMQ connections and handle queue migrations during channel recovery
  • fix: Return raw ClickHouse result tables for trace aggregation and projection queries

🔔 Notifications

  • fix: Prevent Google Chat API calls from blocking the event loop

🎫 Tickets

  • fix: Allow read-only users to create tickets and add ticket comments
  • fix: Normalize scheme-less Jira configuration URLs prior to host validation

🔄 Workflow

  • feat: Support multi-source system runbook templates with browseable labels
  • feat: Add client-side validation for timezone and date-format template filters
  • fix: Resolve issue where approved workflow tasks remained stuck in SCHEDULED state
  • fix: Enable provider-aware validation for Get Ticket workflow forms
  • fix: Improve matching for legacy webhook URLs during workflow publishing
  • fix: Preserve auto-optimize execution history and record failed runs in the Tasks tab

☁️ Cloud

  • feat: Enhance GCP observability with Cloud Run response-code metrics, structured audit log attributes, and deployment change enrichment
  • feat: Support applying and reverting alert threshold suggestions directly or via Pull Request
  • feat: Calculate and display blast-radius safety bands and default impact scoring on workload recommendations
  • feat: Add alternate instance type selectors and fallback resolution for EC2 recommendations
  • fix: Decouple post-report synchronization from spend-fetching failures
  • fix: Capture diagnostic evidence for GCP metric alert events
  • fix: Optimize ECS task metadata fetching and eliminate region-wide scans during EventBridge processing
  • fix: Accurately report stopped GCP Cloud SQL instances as STOPPED
  • fix: Skip AWS regions where services return InvalidAction errors
  • fix: Exclude Azure Defender assessments from resource persistence
  • fix: Correct memory-unit formatting in right-sizing Pull Request creation

🤖 AI

  • feat: Support per-model context windows, o200k token counting, budget-aware recovery, and custom deployment gating
  • feat: Optimize coding agent token usage via run-scoped memory, bounded tools, and incident evidence summarization
  • feat: Enable tenant-wide b-Cortex/Settings reads and streamline memory RPC endpoints
  • feat: Enable @finops agent to propose recommendation applications via UI hand-off
  • feat: Ground automation-builder event triggers in real event schemas and explain proposed edits
  • feat: Add egress filter action gate hooks, per-message metadata tracking, and UI indicators
  • feat: Enable turn-1 parallel fanout for investigations and surface structured triage reasoning in event analysis
  • feat: Enhance AI Cost Analyzer with Tools tabs, Users leaderboards, prompt-trace viewers, severity highlighting, and token cleanup cron jobs
  • feat: Make Nubi follow-up and assistant panels collapsible
  • fix: Enable PR-followup metadata edits on Projects-classic organizations
  • fix: Prevent internal account IDs and provider tags from appearing in workflow-builder follow-up options
  • fix: Automatically build automations on first approval in nested chat flows
  • fix: Improve tool execution reliability with schema-aware input validation and enhanced error hints
  • fix: Prevent AWS investigation agent hallucinations and unsafe command executions
  • fix: Cap log fetching limits to match Loki caps and resolve stale prompt issues
  • fix: Harden background memory jobs and resolve concurrency races
  • fix: Guard against missing pod container resources in cluster right-sizing fallbacks
  • fix: Prevent token-usage metric panics and eliminate double-counting of Gemini cached tokens
  • fix: Gate scratchpad context compression on pressure and display compression cause on visibility cards
  • fix: Gate event analyzer synthesis step on completion and regeneration states
  • fix: Prevent system-user sentinels from forwarding as user IDs and invalidate stale AI cache pointers

🔌 Integrations

  • feat: Support configuring multiple RabbitMQ integrations per account
  • feat: Support per-source webhook subject and namespace label mapping