Skip to content

Changelog

All notable changes to OmniObserve are documented here.

This changelog follows Semantic Versioning and is generated from CHANGELOG.json using schangelog.

v0.6.0 - 2026-02-22

Highlights

  • Built-in slog provider for local trace logging during development and debugging
  • Structured evaluation integration for connecting llmops with sevaluation workflows

Added

  • llmops/slog provider for local structured logging of trace events
  • WithLogger option in ClientOptions for custom slog.Logger configuration
  • integrations/sevaluation package for structured-evaluation integration

Fixed

  • gosec warnings for APIKey struct fields (G117) and HTTP client SSRF (G704)

Documentation

  • Observability integration guide for stats-agent-team
  • README updated with slog provider in supported providers table

v0.5.1 - 2026-01-19

Highlights

  • Compatibility update for omnillm v0.11.0 with new multi-provider configuration API

Changed

  • Updated ClientConfig usage to new Providers slice API for omnillm v0.11.0 compatibility

Dependencies

  • Upgraded github.com/plexusone/omnillm from v0.10.0 to v0.11.0

v0.5.0 - 2026-01-03

Highlights

  • Comprehensive evaluation metrics for assessing LLM outputs with both code-based and LLM-based metrics
  • Streamlined provider architecture with Opik and Phoenix adapters moved to standalone SDKs

Added

  • AnnotationManager interface for span/trace annotations with CreateAnnotation and ListAnnotations methods
  • DatasetManager methods: GetDatasetByID and DeleteDataset
  • Prompt model/provider options: WithPromptModel, WithPromptProvider, ModelName, ModelProvider fields
  • OmniLLM hook auto-creates traces when none exists in context
  • Trace context helpers: contextWithTrace and traceFromContext
  • llmops/metrics package with LLM-based metrics: HallucinationMetric, RelevanceMetric, QACorrectnessMetric, ToxicityMetric
  • llmops/metrics package with code-based metrics: ExactMatchMetric, RegexMetric, ContainsMetric
  • examples/evaluation demonstrating metrics usage

Changed

  • BREAKING: Provider adapters moved to standalone SDKs: Opik to github.com/agentplexus/go-opik/llmops, Phoenix to github.com/agentplexus/go-phoenix/llmops

Removed

  • llmops/opik adapter (moved to go-opik)
  • llmops/phoenix adapter (moved to go-phoenix)
  • sdk/phoenix package (use go-phoenix directly)

v0.4.0 - 2025-12-27

Highlights

  • Full-stack observability with new agentops and observops packages for monitoring agentic AI systems
  • Semantic conventions aligned with OpenTelemetry for standardized telemetry across LLM applications

Added

  • agentops package for agent operations monitoring
  • observops package for unified observability operations
  • semconv package with semantic conventions for agentic AI

Changed

  • llmops refactored to use github.com/plexusone/omnillm

v0.3.0 - 2025-12-21

Highlights

  • Project renamed from observai to omniobserve for consistency with omnillm ecosystem naming

Changed

  • BREAKING: Project renamed from observai to omniobserve for consistency with omnillm naming conventions

v0.2.0 - 2025-12-21

Highlights

  • FluxLLM integration enables observability for flux-based LLM workflows

Added

  • integrations/fluxllm package for FluxLLM observability

v0.1.0 - 2025-12-20

Highlights

  • Initial release providing unified observability primitives for LLM applications with OpenTelemetry integration

Added

  • Core observability framework for LLM applications
  • GitHub Actions CI workflow
  • GitHub Pages presentation
  • MIT License