Find a pattern
Go to Patterns and select a pattern from the list to open it. The list shows open patterns by default. You can close a pattern yourself, and Loop can close one when recent data shows that the behavior has stopped. Click Filter patterns to switch between Open, Closed as resolved, Closed as not useful, and Closed without reason. Each option shows how many patterns it contains. To search across patterns by name, use the search box above the list.Read a pattern
Start here to decide whether a pattern is worth acting on, and to check Loop’s reasoning against the traces it drew from. A pattern detail has four parts:- Summary is what Loop found and the context behind it.
- Suggested fix is the next investigation or change Loop recommends, when it has one to offer.
- How to monitor appears when Loop recommends a scorer, classifier, or facet to measure the behavior. See Act on a pattern.
- Evidence is what the finding rests on.
- Frequency measures how often the behavior occurs when existing trace data can reliably identify it.
- Additional Monitors show complementary measurements, such as latency, cost, or score trends, when Loop attaches them.
- The trace list shows the traces themselves, with Loop’s notes quoted inline. Select a trace to open it in the Debugger. The trace scrolls to the span field the evidence came from and highlights the quoted terms.
Frequency
The Frequency chart shows the percentage and number of traces affected in the selected time range. Each trace counts once, even if several spans match the behavior. Use the time range selector above the chart to measure a different window. When Loop identifies a comparable population, the chart defaults to Similar. This measures the share of comparable traces affected, such as requests for the same operation. Select All to measure the share across your project. The description below the chart identifies the population. If no separate population is defined, the chart shows All without a scope dropdown. Frequency is absent when existing data cannot reliably identify affected traces, for example, when the behavior needs a classifier that has not been created or backfilled. Loop can still support the finding with trace evidence and recommend how to measure it.Act on a pattern
Once you’ve read a pattern, there are four things you can do with it. They’re alternatives, not a sequence, and which one fits depends on whether you want to understand the behavior better, track it, fix it, or bring in someone else.Keep investigating
Select Continue in Loop to start a new thread with the pattern attached as context. Loop shows a badge with the pattern name, so you can see what it’s working from.Make it measurable
Under How to monitor, click Create scorer or Create classifier, depending on what Loop recommended. Loop opens with a prefilled prompt, adds the evaluator to the pattern, works out a backfill strategy with you, and creates a monitor chart for it. This is how a one-time finding becomes something tracked continuously, at lower cost than re-investigating it. Loop can also turn a pattern into a facet so that Topics classifies the behavior across all your traces.Hand it to a coding agent
Click Copy pattern as prompt in the pattern header. Braintrust copies a Markdown brief containing the pattern’s status and close outcome, summary, suggested fix, evidence trace IDs, andbt CLI commands for pulling the full traces. Paste it into your coding agent to start work on a fix.
Share it
Click Copy link in the pattern header to copy a link to the pattern. This appears when the pattern is open in Loop’s workspace. The link only works for people who can already access the project.Objects a scheduled run creates are not linked back to the pattern. If a pattern’s monitor chart measures a scorer or facet that a run created, deleting that scorer or facet stops new data from reaching the chart.
Close or reopen
Close a pattern once you’ve acted on it, or once you’ve decided it isn’t worth acting on. Closed patterns leave the Open list, and the outcome you choose becomes feedback Loop uses to judge future findings. Click Close in the pattern header. Choosing an outcome is optional:- Resolved if the problem is fixed, addressed, or no longer happening.
- Not useful if the pattern is noise, transient, or unimportant.
- Not useful tells Loop not to automatically reopen or recreate a matching pattern while your reason still applies.
- Resolved, or closing without an outcome, allows Loop to reopen the existing pattern if the behavior recurs, unless your feedback says otherwise.
- Free-text feedback defines the scope of your judgment. If you dismiss synthetic staging traffic but new evidence comes from production, Loop evaluates that evidence independently. Be specific about what you want to dismiss and any conditions for reopening.
Update permission on the project. Without it, the Close and Reopen buttons don’t appear, though you can still read patterns and their evidence. Members of the Owners and Engineers permission groups have it by default.
Next steps
- Write scorers to measure a behavior a pattern surfaced.
- Build datasets from the traces behind a pattern, then run an experiment to test a fix.
- Create custom facets to track the behavior across every trace.
- Debug a trace to dig into a single piece of evidence.