v0.8.1

What's Changed

• Durable storage backend (preview). New opt-in Postgres backend for persisting evaluation runs and their results. The Helm chart now ships an optional bundled Postgres for easy trials. APIs and schema may change without notice while this matures. (#135)

Fixed

• Wheel publishing to PyPI works again; a packaging defect produced duplicate file entries that PyPI rejected. (#138)
• PyPI project page now renders the README logos correctly. (#126, thanks @frivas-voiceatlas)

Upgrade notes

• No action required for existing users. The default in-memory backend is unchanged.
• 0.8.0 was yanked due to the packaging issue above. Pin to 0.8.1 or later.

Full Changelog: v0.7.3...v0.8.1

v0.7.3

What's Changed

• order the release pipeline by @peterj in #125
• reference OCI chart location instead of local path in docs by @krisztianfekete in #128
• Add support for traces exported from Tempo by @krisztianfekete in #129

Full Changelog: v0.7.2...v0.7.3

v0.7.2

What's Changed

• publish agentevals helm chart by @peterj in #124

Full Changelog: v0.7.1...v0.7.2

v0.7.1

What's Changed

• feat(examples): add Pydantic AI zero-code OTLP example by @frivas-at-navteca in #117
• Improve trace loading logic by @krisztianfekete in #123

Full Changelog: v0.7.0...v0.7.1

v0.7.0

What's Changed

• Improve nix flake by @krisztianfekete in #111
• Simplify README, update OTel docs by @krisztianfekete in #110
• Use flake-utils.lib.mkApp, keep venv in sync by @krisztianfekete in #112
• Fix/sse request alias by @frivas-at-navteca in #116
• Clean up cli output, fix classification bug by @krisztianfekete in #118
• Chore/adk v1.30 upgrade by @krisztianfekete in #120

New Contributors

• @frivas-at-navteca made their first contribution in #116

Full Changelog: v0.6.4...v0.7.0

v0.6.4

What's Changed

• (docs): Add k8s example by @krisztianfekete in #86
• Expose all tool_trajectory_avg_score match types by @ossama-ferjani in #89
• Add zero code Ollama example by @ajimenez1503 in #91
• npm audit fix by @krisztianfekete in #103
• Improve MCP server by @krisztianfekete in #104
• Update README by @krisztianfekete in #105
• gRPC OTel Receiver support by @ajimenez1503 in #98
• Extend SDK with perf metrics by @krisztianfekete in #108
• Support OTel gen_ai semconv v1.40 by @krisztianfekete in #109
• Create StreamingTraceManager once by @ajimenez1503 in #107

New Contributors

• @ossama-ferjani made their first contribution in #89
• @ajimenez1503 made their first contribution in #91

Full Changelog: v0.6.3...v0.6.4

v0.6.3

What's Changed

• Fix otlp session fragmentation by @krisztianfekete in #83
• fix dropping invocation scores and details and UI by @krisztianfekete in #85

Full Changelog: v0.6.2...v0.6.3

v0.6.2

What's Changed

• Improve session groupping by @krisztianfekete in #82

Full Changelog: v0.6.1...v0.6.2

v0.6.1

What's Changed

• Consolidate trace-to-invocation conversion by @krisztianfekete in #76
• Replace _trace_manager variables with FastAPI dependency injection by @krisztianfekete in #78
• Consolidate text extraction into a single source of truth by @krisztianfekete in #79
• Consolidate duplicated tool call extractions by @krisztianfekete in #80
• add OpenAI Agents SDK zero-code example by @shahar-dagan in #72
• agentevals docker build and helm deploy by @esara in #66

New Contributors

• @shahar-dagan made their first contribution in #72
• @esara made their first contribution in #66

Full Changelog: v0.6.0...v0.6.1

v0.6.0

What's Changed

• Updating the readme.md file by @ProfessorSeb in #64
• Readme add overview sections by @ProfessorSeb in #63
• Add venv support for custom evals by @krisztianfekete in #65
• Add openai_eval type to delegate evals to OpenAI APIs by @krisztianfekete in #73
• document OpenAI Graders by @krisztianfekete in #74
• add per eval time metrics to CLI by @krisztianfekete in #75

New Contributors

• @ProfessorSeb made their first contribution in #64