Honeycomb changelog honeycomb.io

New

Query Math lets you define multi-step queries and perform math across their results directly in the Query Builder, now available across all your datasets. No calculated fields required.

Each step has its own SELECT, WHERE, and GROUP BY clauses, and you combine them using simple formulas like ($A / $B) * 100. Give each formula a name and it carries through to your visualizations, keeping your charts focused on the final calculated value. Common use cases include:

• Error and success rates — count errors in Step A, total requests in Step B, divide and multiply by 100
• Memory utilization — compare memory usage against available memory across your pods
• Request rate differences — spot load imbalances by comparing request rates between services

When you're ready to act on your data, you can alert on formula results directly from a Trigger.

Support for query math is shortly rolling out to Canvas and MCP as well!

New

You can now launch an AI-powered investigation with Canvas directly from a Trigger or SLO Burn Alert.

Trigger and SLO Burn Alert notifications sent to Slack, PagerDuty, and Email now include an ✨ Investigate link. Clicking this link starts an automatic investigation with information gathered about your alert, including trigger or SLO details, thresholds, and datasets.

Use Canvas Alert Investigations to reduce context gathering and get to root cause faster with Honeycomb.

New

You can now add thresholds to time series charts, including: line, stacked area, stat, and bar charts and save them to Boards or your Saved Queries. Thresholds help you quickly see system health and spot when your data cross values that require additional attention or action.

To add a threshold, select Modify Chart from your query results and select a threshold value, color and label. When you save the query to a Board or your Saved Queries, the threshold is saved with it.

New

This GA release brings broader platform coverage and faster debugging for frontend teams.

Highlights

• Readable Stack Traces Across All SDKs: Errors GA delivers human-readable stack traces via Honeycomb’s Symbolicator plugins reducing time spent deciphering minified or opaque error output
• Improved Android Exception Handling: Native Android applications now benefit from symbolicated uncaught exceptions providing clearer insights into runtime failures
• React Native support for Errors & OpenTelemetry SDK compatibility available

New

You can now save filter configurations on Boards as views! Views preserve your exact filter settings so you can return to the precise context you need with a single click, eliminating the need to rebuild the same scope each time.

Begin by defining filters. Once you've assigned specific values to each filter dropdown, select Save as view and name your view. When you next return to the Board, simply select the view you want from the view dropdown. Learn more about views in our docs.

New

Categorical charts have a new setting: the ability to hide the other category data. Other category data can help give a holistic understanding of how different groups compare to the total, but sometimes, you want to focus in on what matters. This is particularly useful on high cardinality fields, where other category data may make up a large percentage of the total.

New

You can now set preset filters on a Board to guide how its data is explored. These filters give every user quick access to the most useful slices of data, ensuring consistent and relevant results, without needing to know which fields matter beforehand.

Learn more in our docs.

Improvement

The SLO Details Heatmap now displays SLI success and failure rates on two separate graphs, and supports hover details and click actions, including the ability to click to a trace.

Improvement

Canvas now supports team-level custom context that is automatically included in every prompt your teammates send through Canvas.

You can use this to provide shared background information, domain-specific details, team conventions, or anything else that should consistently guide Canvas’ reasoning. Once set, the context is seamlessly prepended to all Canvas interactions—no extra steps required.

This helps teams keep responses aligned, reduces repetitive prompt setup, and ensures Canvas always has the “ground truth” your team relies on.

Improvement

You can now specify a SLO Burn Alert in terms of a target Burn Rate for alerting. Burn rates provide another way to intuitively prescribe alerting conditions for your SLOs.

Example:

Specifying a Burn Rate allows you to target a rate of error budget consumption. With the ideal 1x burn rate, the SLO budget is consumed at exactly the end of the SLO window. A 2x burn rate burns through budget in half that time, 3x run rate in 1/3 of the SLO window, etc.

Learn more in Honeycomb Docs