Repository Issue Activity (beta)

#2706 DOC ConversationScorer comments say tool output is excluded from the scored text; the code includes it

#2701 BUG Converter Registry dialogs lose keyboard focus after dismissal and duplicate errors

#2665 MAINT Complete GCG optimization phase extraction after #2416

#2703 FloatScaleThresholdScorer accepts a NaN threshold and then records every response as a complete False verdict

#2710 BUG: ScorerMetrics.to_json() raises TypeError on the trial_scores array ScorerEvaluator attaches

#2265 FEAT Add a WildGuard scorer following the LlamaGuard and ShieldGemma pattern

#2702 BUG Legacy /targets links redirect to Home instead of Target Registry

#2693 BUG Console printers write ANSI/OSC sequences from target responses straight to the terminal

#2683 FEAT: Add ANSI escape output scorers and Garak AnsiEscape scenario

#2535 FEAT: Add Garak API key scenario

#2088 Add a `code_attack` technique variant that applies the code-completion system prompt

#1163 Add GOAT orchestrator

#2672 TEST Validate non-default dataset variants against real upstream artifacts

#2490 BUG Make GCG random_seed deterministic across workers and devices

#2680 Scanner run preview dismissal does not restore keyboard focus

#2489 FEAT Add generic GCG model adapters and an open-weight SLM compatibility matrix

#2661 Scorer evaluation category matching is guarded on the wrong axis and can silently score the wrong category

#2594 Configuration silently discards unsaved changes during navigation

#2647 Add Pinyin Mix converter for Chinese adversarial text transformations

#2641 Selected configuration file path fails contrast in light and dark themes

#2590 BUG: BinaryConverter rejects characters in words the selection strategy never converts

#2570 Scorer evaluation fails when the dataset harm category differs from the harm definition category

#2506 Converter parameter controls are not associated with their visible labels

#2642 Feedback dialog becomes unusable after choosing Go back and fix

#2628 Report harm scorer accuracy split by inter-rater agreement (a third of violence gold labels are single-vote)