• Good over total anomaly detection
• Replay for composite SLOs
• Reporting tools for MCP clients
• SLI Analyzer tools for MCP clients
• Zscaler integration (beta)
• User access key administration

Release details​

new

Good over total anomaly detection​

Good over total anomaly detection checks live data. It identifies intervals where good event counts exceed total event counts in ratio SLOs configured with good and total count metrics, helping you find mismatched queries or inconsistent source data.

For incremental metrics, detection accounts for delayed good events that catch up with an earlier total count within a one-minute window by default.

See Good over total anomaly for detection behavior and troubleshooting guidance.

new

Replay for composite SLOs​

You can now replay composite SLOs to recalculate their error budgets for a selected period using their component SLOs' existing historical data.

Select Include historical data in the Budget step when creating a composite SLO, or choose Run Replay from an existing composite SLO's details page. Use Replay to populate a new composite's history or refresh it after replaying a component, adjusting a component's past error budget, or changing the composite's configuration.

See Replaying composite SLOs for instructions.

new

Reporting tools for MCP clients​

Nobl9 MCP clients can now retrieve resource usage summaries and Reliability Roll-up data for saved reports or ad hoc report definitions. Ad hoc roll-ups are calculated without saving or updating a report.

By default, both roll-up tools return the folder hierarchy and scores without individual SLO entries. Set view: "full" to include individual SLOs.

See the MCP tool list for reporting capabilities.

new

SLI Analyzer tools for MCP clients​

Nobl9 MCP clients can now manage SLI analyses, inspect their status and results, and generate SLO YAML templates from completed calculations. The generated YAML includes placeholders to complete before applying the SLO.

See the MCP tool list for available SLI Analyzer operations.

new

Zscaler integration (beta)​

Connect Zscaler Digital Experience (ZDX) to Nobl9 using an Agent or Direct connection in the Beta release channel. Create threshold SLOs from application ZDX scores and page fetch times, device Web Probe timings and availability, or CloudPath latency and packet loss. The integration also supports historical data retrieval with Replay and SLI Analyzer.

In Nobl9 Web, the Zscaler SLO form provides searchable suggestions for applications, locations, probes, and CloudPath segments. Device ID remains a manual input, and IDs in YAML and API definitions remain integers.

See Add Zscaler data source for setup instructions and Create Zscaler SLOs for supported metrics and query configuration.

new

User access key administration​

The new All user access keys tab in Settings is available only to Organization Admins and lets them view and manage user access keys across the organization. The list shows keys belonging to active users, including inactive and expired keys, with owner details, client IDs, descriptions, statuses, and creation and expiration dates.

Search by owner name, email, or client ID, filter by key status, and sort by creation or expiration date. Export as CSV downloads all keys matching your search and filters. Key secrets are not shown or exported.

Organization Admins can enable, disable, or delete keys. Disabling and deleting require confirmation; expired keys cannot be enabled.

• Replay tools for MCP clients

• Elasticsearch Direct in the Stable release channel

Release details​

new

Replay tools for MCP clients​

Nobl9 MCP clients can now work with Replay workflows directly. They can check whether Replay is available, start a Replay, monitor its status, list active Replays, cancel eligible runs, and delete queued Replay requests for an SLO.

improved

Elasticsearch Direct in the Stable release channel​

Elasticsearch Direct data sources can now be created in the Stable release channel. Stable Elasticsearch Direct data sources can use historical data retrieval features such as Replay and SLI Analyzer.

• API key creator and change history

Release details​

new

API key creator and change history​

API key details now show who created the key under Created by, and a Change history tab lists what happened to the key afterwards: creation, edits to the display name and description, activation and deactivation, and organization and project role changes. You can filter the list by event type and tool.

Change history is available to Nobl9 Enterprise customers, on the same terms as change history for SLOs, projects, and services.

• Access keys can only be managed by their owner
• Single-step Codex setup for the Nobl9 MCP server
• Samples at a midnight range end in the Prometheus API

Release details​

fixed

Access keys can only be managed by their owner​

Fixed the handling of access keys that belong to another user or to an agent. Knowing a client ID was enough to delete, deactivate, activate, or rename such a key. You can now manage only the access keys you own.

fixed

Single-step Codex setup for the Nobl9 MCP server​

Connecting Codex to the Nobl9 MCP server is now a single step. Running codex mcp add starts the Nobl9 sign-in itself, so the separate codex mcp login step is no longer required. The updated instructions are available under Integrations > Connectors.

fixed

Samples at a midnight range end in the Prometheus API​

Fixed range queries in the Prometheus API for SLO time series whose end parameter fell exactly on midnight UTC. The last data point returned a stale value from the previous day or was missing from the response. The end timestamp is now inclusive, matching the Prometheus API contract.

• Easier AI client connections to the Nobl9 MCP server
• Reference Nobl9 resources in Assist

• Compare installed and latest Agent versions

• Warnings for incomplete SLO time series results

Release details​

new

Easier AI client connections to the Nobl9 MCP server​

You can now connect compatible AI clients, including Claude Code, Cursor, Codex, and more, to the Nobl9 MCP server without providing a client ID. Compatible clients register automatically and open the Nobl9 sign-in page. Setup instructions for supported clients are available under Integrations > Connectors.

new

Reference Nobl9 resources in Assist​

In Nobl9 Assist, type @ to find and reference a project, service, SLO, or data source in your prompt. Selected resources appear as pills in the composer and sent messages, helping Nobl9 Assist identify the exact resources you mean.

improved

Compare installed and latest Agent versions​

Agent data source details now show Installed version and, when a newer version is available, Latest version for the configured release channel. You can compare both versions at a glance.

fixed

Warnings for incomplete SLO time series results​

When Nobl9 cannot read all metric data, the Prometheus API for SLO time series returns the available data with a warning that the result may be incomplete. The warning recommends retrying the request and provides a request ID to share with Nobl9 support if the problem persists.

• Data sources in Global Search

Release details​

improved

Data sources in Global Search​

Global Search (Ctrl+K) now also finds data sources. Type a data source name or display name to jump straight to its details page.

• Meet Nobl9 Assist

Release details​

new

Meet Nobl9 Assist​

Nobl9 Assist is an in-context AI assistant in Nobl9 Web. Ask in plain language which SLOs are burning, why a budget is draining, or how to configure a feature, and it answers with live data from your organization. It can also draft and apply configuration changes, but only after your explicit approval in the chat.

Starting with this release, users with access to Nobl9 Assist see an in-app announcement with a guided tour.

• Connectors tab on the Integrations page
• Elasticsearch Direct support

Release details​

new

Connectors tab on the Integrations page​

The Integrations page now includes a Connectors tab with setup guidance for tools that connect Nobl9 to external workflows.

The tab includes connector options for the Nobl9 MCP server, APIs, and configuration tools such as sloctl, Terraform, and the Go SDK.

new

Elasticsearch Direct support​

Elasticsearch Direct is now available in the Beta release channel.

You can use Elasticsearch Direct for live SLI ingestion, Test metric, Replay, and SLI Analyzer workflows with encoded API key authentication.

• Count delta metrics in the Prometheus API

• Expandable annotation window
• Multiple active disruptions per component
• Release update indicator in Help & Resources

• Anchored regex label matchers in the Prometheus API

Release details​

new

Count delta metrics in the Prometheus API​

The Prometheus API for SLO time series now exposes count_good_delta and count_total_delta, which report the increment since the previous sample. Both are also available pre-aggregated at 1-minute, 5-minute, 15-minute, 1-hour, and 1-day resolutions.

Unlike count_good and count_total, which cover the whole SLO time window and can decrease as older observations leave it, the delta metrics report fresh increments only. This makes them a better fit for analyzing event volume over time. See the count delta metrics reference for details.

improved

Expandable annotation window​

You can now expand an annotation popover into a larger, viewport-centered window with the new expand control next to Close. The header and Description section stay visible while the annotation content scrolls independently, making long descriptions easier to read and edit.

improved

Multiple active disruptions per component​

The disruption API now allows multiple active disruptions to share the same origin component. Previously, a component could have only one impacting disruption at a time.

The component's status continues to reflect the most severe active disruption. The activeDisruptionId field in API responses now refers to the newest active disruption.

improved

Release update indicator in Help & Resources​

The Help & Resources menu now shows a notification dot when new Nobl9 release notes are available, so you can spot product updates without leaving the app.

fixed

Anchored regex label matchers in the Prometheus API​

Regex label matchers in Prometheus API queries are now anchored, matching the behavior of native Prometheus.

Previously, a selector such as slo=~"name-a|name-b" could also match SLOs whose names merely contained one of the alternatives. This over-matching could inflate a query's series count and fail the query with a query selects N series, which exceeds the maximum error.

• Replay events in SLO Change History

• Clearer Replay and adjustment elapsed time
• Expanded Test metric support
• Faster processing for large Replay backfills

Release details​

new

Replay events in SLO Change History​

Replay activity now appears in the SLO Change History. You can see when a Replay was requested, started, completed, canceled, extended, deleted, or failed.

These entries include the user or system actor behind the change, the tool used to request it, and the Replay state transition. This makes it easier to understand why an SLO's error budget changed after historical data was reprocessed.

improved

Clearer Replay and adjustment elapsed time​

The Replay and adjustment job status now measures Elapsed time from the moment processing starts, instead of including time spent waiting in the queue.

This makes the job status widget easier to interpret during busy periods. The same timing behavior is also reflected in Replay list data used by sloctl replay list.

improved

Expanded Test metric support​

Test metric now supports more data sources. With Nobl9 agent 0.114.0-beta or later, you can preview raw SLI data for the following:

• Amazon Prometheus
• Azure Monitor managed service for Prometheus
• Elasticsearch
• Google Cloud Monitoring
• LogicMonitor
• Prometheus
• ThousandEyes

improved

Faster processing for large Replay backfills​

Large batches of queued Replays now start at a steadier rate. Previously, a very large Replay backlog could slow down queue processing, leaving available concurrent Replay slots unused and extending the total backfill time.

The queue now scales better when many Replays are waiting, so large historical data backfills can keep moving more consistently.

• Elasticsearch Agent support in SLI Analyzer

• Error-budget anomalies after objective changes

Release details​

new

Elasticsearch Agent support in SLI Analyzer​

You can now use eligible Elasticsearch Agent data sources in SLI Analyzer.

fixed

Error-budget anomalies after objective changes​

Fixed an issue where an active error-budget anomaly could stay in the Detected state after you changed an SLO objective value.

Nobl9 now cancels the previous anomaly detector state when the objective value changes, so existing anomalies are resolved and the updated objective starts with fresh anomaly detection state.

• Webhook HTTP signatures
• Treat zero values as no data
• Splunk Observability generally available
• Disruption timeline updates

Release details​

new

Webhook HTTP signatures​

Nobl9 now signs all webhook notifications with HTTP Message Signatures.

Use the signature to verify that a webhook notification came from Nobl9 and that the signed request fields and body were not changed in transit. Signed requests carry the Content-Digest, Signature-Input, and Signature headers.

Existing webhook endpoints continue to receive notifications without changes. Signature verification is optional, and you can adopt it when your endpoint is ready.

new

Treat zero values as no data​

No data anomaly detection can now treat a value of 0 as no data, for data sources that return 0 as a placeholder instead of omitting the data point.

Enable the setting per SLO to have zero values trigger no-data alerts. Existing SLOs keep the previous behavior until you enable it.

new

Splunk Observability generally available​

The Splunk Observability integration is now self-service for all organizations.

new

Disruption timeline updates​

Disruptions & impact in Organization status now uses a new update flow for posting changes to a disruption timeline.

A single update can add a timeline comment, change the severity, and amend the scope of impacted components.

• Disruption wall
• Test metric in the SLO wizard
• MCP data source queries

Release details​

new

Disruption wall​

Organization status now includes the Disruption wall, a dedicated view for ongoing disruptions in the major outage status. Disruption cards show origin context, affected components, updates, and available actions in one place.

new

Test metric in the SLO wizard​

The SLO wizard now includes Test metric for supported data sources. Use it to preview raw SLI data for the last 15 minutes or 1 hour before saving an SLO.

Test metric opens a read-only chart with returned values and timestamps. If the data source returns an error, the wizard shows validation details so you can adjust the metric before continuing.

new

MCP data source queries​

The Nobl9 MCP server now exposes the queryDataSource tool. AI clients connected to Nobl9 can use it to query supported data sources for time-series data over a requested time range.

The response includes returned timestamps and values. When a data source rejects a query, Nobl9 returns the same validation context used by other interactive query flows.

• Resource center
• Event Logs for Splunk Observability
• Global project scope for reports
• MCP tools for Organization status
• MCP documentation search tool

• Splunk Observability query validation

Release details​

new

Resource center​

The Resource center is available in the top navigation and gives users quick access to Nobl9 announcements and feature guides.

new

Event Logs for Splunk Observability​

Event Logs are now available for Splunk Observability direct data sources.

The Logs tab and Activate logs action are available for Splunk Observability data sources, and the API accepts enabling log collection for them.

new

Global project scope for reports​

Reports can now be scoped to all current and future projects.

Users with organization-wide project access can select a global project scope so a report automatically includes every project in the organization, including projects created after the report is configured.

Learn more in Creating reports.

new

MCP tools for Organization status​

The Nobl9 MCP server now exposes read-only tools for Organization status information:

• getOrganizationStatus returns the component tree with computed status and ongoing disruptions.
• listStatusPageDisruptions returns open and cleared disruptions.

MCP clients can use these tools to answer status questions such as whether anything is currently down for an organization.

new

MCP documentation search tool​

The Nobl9 MCP server now exposes a read-only searchDocs tool.

MCP clients can search the official Nobl9 documentation and return matching pages with titles, links, and snippets, helping assistants answer how-to and conceptual questions with links to docs.nobl9.com.

improved

Splunk Observability query validation​

The SLO query checker now validates Splunk Observability queries on the beta release channel.

Invalid SignalFlow programs are reported on the SLO Details screen with the error returned by Splunk Observability. Splunk Observability is also available to every organization by default.

• Multi-component Status Page disruptions
• Manage Status Page disruptions over HTTP API

Release details​

new

Multi-component Status Page disruptions​

You can now register a single Status Page disruption on multiple origin components at once, including components from different parts of the component tree.

The disruption propagates to the ancestors of every origin, and a single clear, update, or severity change then applies to all of them.

new

Manage Status Page disruptions over HTTP API​

Machine-to-machine integrations can now manage Status Page disruptions using HTTP Basic authentication with an access key (clientId:clientSecret), the same method already used for external issue reporting.

External monitoring and automation systems can register a disruption, change its severity, and clear it without a browser session.

• Empty-state experience updates

• Status Page component selector performance

Release details​

improved

Empty-state experience updates​

Empty states now provide a more consistent experience across SLOs, Catalog, SLI Analyzer, Integrations, Status Page, and dashboard views.

fixed

Status Page component selector performance​

The Status Page component selector is now faster and more responsive when you create incidents for status pages with large component sets.

• Prometheus API for SLO time series
• Dynatrace DQL support
• Generate ad-hoc Reliability Roll-up report API
• MCP support for Alert policy analyzer
• MCP support for Prometheus API

• Alert policy analyzer policy selector grouping
• Time window metadata in Alert policy analyzer

Release details​

new

Prometheus API for SLO time series​

The Prometheus API for SLO time series is now available, exposing Nobl9 SLO data through a Prometheus-compatible API.

Use PromQL to query reliability, remaining error budget, burn rate, good and total counts, raw SLI values, and composite SLO component metrics.

Use it when you want to analyze SLO health outside the Nobl9 app: build Grafana dashboards, compare SLOs across projects with PromQL, pull SLO time series into scripts, or feed SLO values into internal reporting workflows.

For example, you can find objectives in a project with less than 20% error budget remaining:

budget{project="my-project"} < 0.2

To get started, see the Prometheus API guide and the Prometheus API reference.

new

Dynatrace DQL support​

Dynatrace DQL query support is now available in the SLO Wizard for the Dynatrace direct data source.

You can author an SLO metric with either the existing Metric selector or a DQL query.

new

Generate ad-hoc Reliability Roll-up report API​

You can now generate Reliability Roll-up report data from a Report manifest without creating a saved report.

new

MCP support for Alert policy analyzer​

The MCP server now supports Alert policy analyzer workflows through tools for calculating alert policy suggestions, starting analyses, fetching results, and retrying failed analyses through the SDK.

It also includes a calculate-alert-policy prompt that helps compare candidate policies for a specific SLO objective and time range and returns analysis URLs for review.

new

MCP support for Prometheus API​

The MCP server now supports Prometheus API for SLOs endpoints for querying SLO metrics data.

improved

Alert policy analyzer policy selector grouping​

The Alert policy analyzer existing-policy selector now groups alert policies linked to the analyzed SLO separately from all other policies. This makes it easier to identify which policy was used to trigger the analyzed alert.

improved

Time window metadata in Alert policy analyzer​

Alert policy analyzer now displays the SLO time window type, so you can see whether the analyzed SLO uses a calendar-aligned or rolling time window.

• Reliability Roll-up reports as code

• Retry failed alert policy analyses

• Splunk Observability stable setup

Release details​

new

Reliability Roll-up reports as code​

Reliability Roll-up reports can now be managed programmatically through the same configuration workflows as other report types. See the kind: Report YAML guide for the Reliability Roll-up manifest structure.

You can create, read, update, and delete Reliability Roll-up reports with sloctl, the REST API, and the Terraform provider. Existing reports created in Nobl9 Web are also returned by sloctl get, including YAML output.

The View YAML action in Nobl9 Web now produces a Reliability Roll-up report manifest that you can re-apply through sloctl.

improved

Retry failed alert policy analyses​

The alert policy analyzer now lets you retry a failed analysis from the analysis result. Use Retry when an analysis fails because of a temporary Nobl9 issue or transient data retrieval problem.

Retrying keeps the same analysis context and attempts to continue the analysis without requiring you to start from the beginning.

fixed

Splunk Observability stable setup​

The Splunk Observability setup now hides historical data retrieval fields when the data source release channel is set to Stable.

Replay and related historical retrieval fields remain visible for the Beta release channel.

• Alert policy analyzer

• Splunk Observability Replay and SLI Analyzer support

• SLO History report performance

• Replay performance

Release details​

new

Alert policy analyzer​

Alert policy analyzer is now available as a Nobl9 Labs feature. It simulates how an alert policy would have behaved against an SLO's historical data before the policy goes live.

Use it to tune alert conditions, validate alert sensitivity, reduce noisy alerts, and compare policy behavior against real budget changes from a selected historical interval.

Start an analysis from SLO details, choose the objective and historical interval, then define the alert policy conditions to test. You can use Auto-fill from time range to generate an initial condition from the budget drop observed over the selected interval and adjust it before running the simulation. After the simulation completes, review the detected alerts, error budget remaining chart, and per-condition charts to decide whether the policy is ready or needs adjustment.

new

Splunk Observability Replay and SLI Analyzer support​

Splunk Observability now supports Replay and SLI Analyzer through a new bounded HTTP-based SignalFlow collection path.

This support applies to newly created Splunk Observability sources. Existing WebSocket-based sources keep their current behavior. The same collection path now powers live data collection for new sources and historical retrieval, aligning live and historical data behavior.

improved

SLO History report performance​

SLO History reports now load large views more efficiently. The update reduces duplicate API calls and improves responsiveness for reports with many SLOs.

improved

Replay performance​

Replay now handles time-series writes more efficiently during the overwriting phase, improving throughput under high load.

• Apply, delete, and validate tools in Nobl9 MCP server

• Error Budget Remaining preview in Reliability Roll-up reports

Release details​

new

Apply, delete, and validate tools in Nobl9 MCP server​

The Nobl9 MCP server now exposes new tools for managing Nobl9 manifest objects from AI clients:

All three tools respect the organization selected in the MCP context, including for multi-organization users.

improved

Error Budget Remaining preview in Reliability Roll-up reports​

The SLO hover preview in Reliability Roll-up reports now shows the Error Budget Remaining chart instead of Reliability Burn Down, aligning the preview with the chart most teams use to track current SLO health at a glance.