Repository Issue Activity (beta)

#2700 Human review queues for low-score, high-latency, and high-cost traces

#2702 Curate production traces into persisted evaluation datasets

#2701 Immutable DatasetVersion snapshots and Run provenance

#2718 [BUG] ruff pinned to three different versions; documented make lint/make format rewrite 380 files

#2704 Portable semantic trace search and failure clustering cookbook

#2442 Streaming LLM Response Support: Token Tracking, TTFT, Partial Output Capture

#2699 [FEAT] `additional_instructions` and `_with_cot_reasons` Support for Conversation-Level Metrics

#2705 Three-step notebook for debugging recurring trace failures

#2703 Connector-backed prompt management with version evaluation

#2680 [FEAT] EvalPort import/export for batch eval datasets and Run results

#2405 Add Ollama Provider (Local Models)

#2619 Leaderboard: trends, significance, drift, alerting (all apps by default)

#2682 [BUG] eval metadata is written namespaced and read back as a bare key, so meta["metadata"] is always empty

#2621 GenAI content as span events + opt-in capture policy

#2440 Nested App / Sub-Agent Tracing (NESTED_RECORD_ROOT)

#2441 Async-Safe Recording: TaskGroup, asyncio.gather, Concurrent Spans

#2470 Add HuggingFace Local Eval Cookbook

#2481 CI/CD Example: GitLab CI Eval Gate

#2627 Generic LLM-call auto-instrumentation

#2444 Batch/Offline Evaluation Mode: Evaluate Datasets Without a Live App

#2640 [FEAT] Add a healthcare administrative evaluation usecase

#2469 Add Bedrock Provider Quickstart Notebook

#2445 CrewAI Integration: Auto-Instrument Agent Crews

#2420 DSPy Integration — Use TruLens Metric as DSPy Metrics

#2622 OTLP-native export + OTel metrics signal