When should I use this skill?

Use it when a user's message triggers the wrong workflow step, when Q&A is ignored, or when 'out-of-context' drops are happening incorrectly.

Does this skill help with testing?

Yes, it provides guidance for adding regression tests in tests/detection/ to ensure that changes to keywords or intent guards don't cause future regressions.

How does it handle the conflict between LLM and Regex signals?

It provides a workflow to compare UnifiedDetectionResult signals from the LLM against step-specific keyword classifiers, helping you decide which signal should take priority for a specific intent.

What does the oe-detection-triage skill do?

It helps developers debug and fix misclassifications in OpenEvent-AI, specifically focusing on intent detection, routing, and reconciling LLM results with regex fallbacks.

How do I enable debug mode for detection triage?

You can run your environment with WF_DEBUG_STATE=1 and set DETECTION_MODE=unified to see detailed logs of the unified detection engine.

OpenEvent Detection Triage

Name: OpenEvent Detection Triage
Author: shami-ah

byshami-ah

Analíticas y Monitorización

Debugs and resolves intent classification, routing errors, and detection misfires within the OpenEvent-AI workflow.

Acerca de

This skill provides specialized guidance for troubleshooting the detection engine in OpenEvent-AI, helping developers reconcile LLM and regex signals to fix routing errors. It streamlines the triage process for misclassified intents, out-of-context (OOC) drops, and unexpected triggers—such as site visits or date changes—by providing step-by-step workflows for log analysis, state debugging, and regression testing. It is essential for maintaining the accuracy of the automated event planning pipeline.

Características Principales

Debugging out-of-context (OOC) evidence gating
Automated trace scripts for deterministic scenario testing
Unified detection signal analysis (LLM vs Regex)
Rapid navigation to routing and classification triggers
Regression test integration for detection logic
0 GitHub stars

Casos de Uso

Investigating why specific Q&A types or confirmation keywords are being ignored
Correcting misclassified user intents that lead to the wrong workflow step
Fixing unexpected site-visit or date-change triggers during user interactions

Acerca de

Características Principales

Debugging out-of-context (OOC) evidence gating
Automated trace scripts for deterministic scenario testing
Unified detection signal analysis (LLM vs Regex)
Rapid navigation to routing and classification triggers
Regression test integration for detection logic
0 GitHub stars

Casos de Uso

Investigating why specific Q&A types or confirmation keywords are being ignored
Correcting misclassified user intents that lead to the wrong workflow step
Fixing unexpected site-visit or date-change triggers during user interactions