2026-02-09 to 2026-02-15

Prev Next

AI / Insights

Features

  • AI Advisor MCP Server: Introduced a new Model Context Protocol (MCP) server that enables Claude Desktop and other AI assistants to directly query Kentik network data. You can now use Claude Desktop to search and filter network devices by various attributes (name, status, site, vendor, model, IP, OS, serial number) through natural language prompts.
  • AI Advisor SavedView Tool: AI Advisor now lets you interact with your saved views through natural conversation — search for saved views by name, execute them directly to instantly retrieve network data, and filter results by ownership and sharing level.
  • Synthetic AI Subtest Results: You can now view synthetic test subtest results through the AI Advisor, including detailed results for page load, URL, hostname, and transaction tests. Also fixes an issue where multiple detail panes in a single session showed duplicate test results.
  • Dynamic Rate Limiting for AI Models: Implemented dynamic rate limiting for Copilot and Claude AI models, ensuring proper API request rate management when using AI assistants for network analysis.

Bug Fixes

  • Alerts Tool Display Fix: Fixed the AI Advisor chat interface so that when interacting with the AlertsTool, the proper alerts results view is displayed instead of a generic default view.

Alerting

Features

  • Public Audit API: Introduced new public audit API endpoints supporting list and get operations, enabling you to access audit information programmatically.
  • Notification Event Retrieval: The Notify API now supports retrieving and test-rendering notification events, allowing you to preview and validate how alert notifications will appear before they are sent.
  • Runbook Active Alert Limit Increase: Increased the active alert count limit in runbook policy assignments from 10 to 2,000 alerts to handle larger datasets.
  • Notification Channel ID in Test Payloads: Enhanced notification testing and preview by appending channel ID information to test and preview payloads, improving logging and tracing.

Bug Fixes

  • Default UA Alert Template Fix: Corrected the default User Analytics (UA) alert policy template which was not functioning properly, ensuring default alert templates work as expected.
  • Runbook Deletion Fix: Fixed runbook deletion failure for runbooks without mappings by ensuring all revisions are properly deleted.
  • Notification Test Domain Fix: Corrected an issue where the base domain was not properly propagated to test notifications, ensuring test notifications are sent to the correct domain.

Cloud

Features

  • OCI Bucket Sweep Simplification: Made the OCI ServiceConnector OCID property optional in Cloud Config, allowing bucket sweep operations to proceed without requiring this field.

Bug Fixes

  • GCP Flow Data in Utilization Chart: Fixed an issue where GCP flow data was not appearing in the Public Clouds Utilization Chart on the cloud settings page.

Core

Features

  • Device Count in Site Table: You can now see a count of devices associated with each site directly in the site table on the Settings > Sites page. The count is clickable and filters the Devices table to show only devices from that site.
  • Bulk Device Addition via Discovery: Added functionality to bulk add devices through the Discovery feature in the Device wizard, enabling you to add multiple devices at once.
  • SNMP Unification Feature Flag Removal: Universal Agent Flow Enrichment and SNMP collection features are now always enabled by default for all users. Device configuration forms are simplified with consolidated logic and improved field validation.
  • IPFIX Sampling Rate Auto-Detection for Juniper: Enables automatic detection of sampling rates directly from IPFIX flow templates on Juniper devices, dynamically using the correct sampling rate for each interface instead of a single fixed rate.
  • Sampling Configuration Management UI: New web-based interface for managing sampling configurations across devices. You can now list, update, delete, and test sampling configurations directly through the UI.
  • NetBox Capability in Universal Agent: Automated collection of NetBox infrastructure metadata (sites, racks, device information) directly through the Universal Agent with daily synchronization intervals.
  • New DNS Services: Added support for 10 new services including BritBox, Ashes of Creation, Clash Royale, Atmosphere.tv, VRChat, Zillow, GitHub Co-Pilot, Microsoft 365 Chat, Microsoft Co-Pilot, and Amazon Q Developer.
  • Traffic Cost Permission Controls: Disabled certain traffic cost slicing options when users lack the required permissions, improving security and compliance.

Bug Fixes

  • Device Status Display Fix: Fixed the device status (NMS) display on the Device Details page where the NMS status was showing as undefined.
  • NMS Device Status Accuracy: Ensured that only devices with full NMS monitoring capability show an NMS status in the device list. Flow-only devices with NMS enrichment no longer display misleading NMS status.
  • Metrics Device Name Filter Fix: Removed an outdated filter clause from kmetrics queries that was preventing results from being returned when filtering by device name.
  • AS Groups Populator Fix: Fixed missing source/destination populators in AS Groups by implementing automatic detection and creation of missing populators.
  • Global Search Navigation Fix: Fixed a critical bug where using keyboard Enter key to navigate in Global Search would result in malformed page paths.
  • Alert Count on Device Details: Fixed a bug where the alert count displayed on the Alert tab of Device Details was rendering incorrectly.
  • Remove Unsupported Route Prefix+Len Filter: Removed a broken combined filter option that was causing errors, directing users to use the separate prefix and length filters.
  • IPv6 Interface Address Fix: Corrected the IPv6 interface address dimension to use the proper IP dimension type instead of IP-prefix, fixing issues where IPv6 interface information was not populating correctly for NMS-polled devices.
  • IPv4 Address Format Fix: Updated IPv4 addresses to use IP address format instead of IP-prefix format, improving consistency in address handling across the metrics system.
  • SNMP Auto-Reconnect on Engine ID Change: Implemented automatic SNMP session recreation when SNMPv3 engine IDs change on devices (e.g., after restart or replacement), preventing permanent loss of monitoring data.

Network Monitoring System (NMS)

Features

  • Device Details Unification: Unified the Device Details Quick View across NMS and Flow interfaces, providing a consistent and streamlined experience when viewing device information across different data sources.
  • Infoblox Model Reclassification: Updated Infoblox device model classifications based on latest MIB information for improved device type accuracy.
  • IPv4/IPv6 Address Report Renaming: Renamed address reports for clarity — oc-address-ipv4 is now the standard report using modern IP-MIB, with -legacy suffix for deprecated versions. Same for IPv6 reports.
  • Infoblox SNMP Trap Template Improvements: Updated Infoblox trap templates to use consistent field mappings and corrected severity field format, ensuring trap data is captured consistently.

Bug Fixes

  • BGP Query Optimization: Optimized the BGP query on the Device Details Overview page by removing an inefficient subquery component, improving query performance.

Synthetics

Features

  • Trace Test Interaction Improvement: Changed synthetic trace test results from hover-based to click-based selection for viewing detailed trace data, with more granular on-demand data fetching using narrower 60-second time windows.

Bug Fixes

  • Target IPs Filter Fix: Fixed the target_ips filter in synthetic test API responses that was previously non-functional, enabling you to correctly filter and retrieve synthetic test results by target IP addresses.
  • Session Deauthorization Consistency: Fixed inconsistent session deauthorization behavior across multiple syn-back instances by using Redis as the single source of truth, preventing agents from operating with stale sessions.
  • Agent Deauth on Task Count Mismatch: Stopped forcibly disconnecting healthy agents when their running task count differs from expected count, which was causing dropped results and 403 errors.