Skip to main content

v1.6.0

What's Changed​

🔍 Troubleshoot​

  • feat: Add topology-inferred blast radius visualization and correlated-incident panel in incident triage
  • feat: Group incident alerts into cause, impact, and noise categories for clearer root cause isolation
  • feat: Support Kibana Query Language (KQL) in Elasticsearch log search
  • feat: Add Elasticsearch and Kibana alerting webhook integrations with automated event closure and subject extraction
  • feat: Expand Knowledge Graph with GitHub, GitLab, and PagerDuty entity sources
  • feat: Add AWS VPC Flow Logs as a network flow source in Knowledge Graph
  • feat: Support nested specific-type filtering under Node Type in Knowledge Graph
  • feat: Disambiguate Knowledge Graph node filter rows by cluster and specific type
  • feat: Integrate service ownership and user groups into the Knowledge Graph topology
  • feat: Add span-level evidence to trace-derived Knowledge Graph nodes and edges
  • fix: Automatically purge inactive Knowledge Graph nodes and reclaim orphaned edges after retention period
  • fix: Derive eBPF trace source accurately from scope names
  • fix: Isolate metric label fetching per query block in telemetry graphs
  • fix: Display exact snooze expiration timestamps in grouped event tooltips
  • fix: Improve diagnostic feedback for unrecognized label names in telemetry queries

⚡ Optimization​

  • feat: Add account picker filter across Auto-Optimize tabs
  • fix: Prevent Auto-Optimize tasks from wedging in Scheduled status
  • fix: Round estimated cost savings values in spot recommendation evidence

🔄 Workflow​

  • feat: Add RabbitMQ API workspace shim for scoped management API task execution
  • fix: Encrypt Temporal workflow payloads and redact sensitive secrets from workflow execution logs
  • fix: Set default approval type for workflow approval tasks so templates load cleanly

☸️ Kubernetes​

  • feat: Introduce K8s-native lean orchestrator mode with reduced tool overhead
  • fix: Support case-insensitive HTTP Basic auth headers in Kubernetes agent middleware
  • fix: Validate Kubernetes namespaces in remediation actions to prevent command injection

☁️ Cloud​

  • feat: Add real-time credential validation when editing GCP billing configurations
  • fix: Skip unroutable AWS regions to keep cloud resource synchronization within execution budgets
  • fix: Automatically terminate orphaned Azure CLI processes upon command timeout
  • fix: Mask authorization headers in failed cloud collector authentication logs

🤖 AI​

  • feat: Add AWS Bedrock support to the AI Gateway with generic endpoints, IRSA, and tenant BYO credentials (bare model IDs are supported for on-demand inference within the same region, with inference profiles recommended for cross-region routing)
  • feat: Add Azure OpenAI support on the AI Gateway with generic endpoints and tenant BYO credentials
  • feat: Add HuggingFace provider support to the AI Gateway
  • feat: Support arbitrary OpenAI-compatible API providers via AI Gateway generic endpoints
  • feat: Add support for self-hosted OpenAI-compatible models (Ollama, vLLM, SGLang)
  • feat: Introduce model tier aliases (fast, cheap, smart) with admin UI for tier mappings
  • feat: Expand AI Gateway Sessions tab featuring per-conversation rollups, search filters, and detailed audit inspection
  • feat: Add prompt cache hit rate and estimated cost savings KPI cards
  • feat: Add per-tenant egress filter admin settings with configurable detection modes and custom regex patterns
  • feat: Surface structured remediation recommendations with runnable command actions on incident event pages
  • feat: Support per-role model tiering for code analysis agents
  • feat: Enhance AI memory retrieval with relevancy-scored reranking and combined-score filtering
  • feat: Enable AI event analysis to inspect unified incident assemblies directly
  • feat: Surface Knowledge Base runbooks in event investigations with cited references
  • fix: Stabilize Claude log investigations, support Claude 5 adaptive reasoning, restore Anthropic prompt caching, and strip unsupported temperature parameters
  • fix: Deduplicate AI memory facts across all storage layers, enable reviewing and reverting inferred user preferences, and scope RCA feedback to message subjects
  • fix: Lift hardcoded output token limits for newer Claude and OpenAI models and improve proactive sub-agent skill loading
  • fix: Ensure Elasticsearch log filters and multi-word searches match accurately, merging schema fields across index patterns
  • fix: Enhance LLM Analyser with provider-grouped account filters, reconciled cost/token metrics, and server-side conversation paging
  • fix: Gate LLM Analyser access via per-tenant feature flags instead of deployment environment variables
  • fix: Enforce per-provider time-to-first-token timeouts and prevent log agent stalls
  • fix: Prevent provider configuration leaks during cross-provider connectivity checks
  • fix: Forward resolved tenant LLM configurations across PR follow-up analyses and code analysis workflows

🔔 Notifications​

  • feat: Introduce Slack channel watching to maintain conversational context and ground AI responses in ongoing channel discussions
  • fix: Preserve anomaly metrics and statistics on event labels for Slack alert notifications
  • fix: Clarify alert suppression wording and link undo actions directly to incident triage rules
  • fix: Route asynchronous LLM error callbacks directly into the initiating conversation thread
  • fix: Prevent crashes during Slack event handling and ensure URLs in notification text are preserved

🎫 Tickets​

  • fix: Scope ticketing search to tenant boundaries and assign unique execution references to standalone task runs
  • fix: Optimize ticket configuration retrieval by replacing N+1 queries with a single batch query

🔌 Integrations​

  • feat: Support Confluence Data Center integration via Personal Access Tokens (PAT) and Basic authentication
  • feat: Migrate webhook subject mappings to dedicated database tables for improved query performance
  • feat: Add Automations and Ask Nubi entry points directly to global page search
  • fix: Enforce strict HTTP client timeouts across integration webhooks to prevent hanging connections
  • fix: Bound database query timeouts and eliminate connection pool deadlocks during integration listing and event filtering