Skip to content

CHANGELOG.md

Latest commit

 

History

History
64 lines (42 loc) · 3.96 KB

CHANGELOG.md

File metadata and controls

64 lines (42 loc) · 3.96 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

1.2.0 - 2026-08-20

Added

  • Microsoft Entra ID authentication for Azure OpenAI via a service principal, selected with auth_method = "entra_id" in llm_config. Credentials are read from AZURE_TENANT_ID, AZURE_CLIENT_ID and AZURE_CLIENT_SECRET, with per-project overrides using the {PROJECT}_ prefix. A project that sets none of the three variables uses the global service principal; setting only some of them is rejected with Entra ID authentication ... is incompletely configured, which names the variables that are missing.
  • azure-identity and aiohttp as runtime dependencies. Both are required for auth_method = "entra_id" and are bundled in the released binary; they are imported only when Entra ID authentication is actually configured, so API key installations are unaffected.

Fixed

  • Azure OpenAI client creation no longer overrides the provider resolved from the model name, so a claude-* model in a prompt variant correctly creates an Anthropic client even when provider = "azure_openai" is configured.

1.1.0 - 2026-07-30

Added

  • A minimum TestBench version check in the precheck of every built-in agent. All three agents require TestBench 4.1 or newer; against an older server the agent now stops immediately with Your current TestBench version ... is not supported, in German or English, instead of failing later with an unrelated error.
  • The Defect Explainer rejects XML-based test object versions in its precheck with The AI agent does not support XML-based test object versions. A TOV whose exchange format is inherited is judged by the project default, so only projects that actually store JSON are accepted.

Changed

  • The Defect Explainer reads the comment format written by testbench2robotframework 2.x, which marks the relevant table cells with data-tb-* attributes. Comments written by tb2rf 1.1 and older are still recognized, so existing cycles keep working.
  • testbench-cli-reporter and testbench2robotframework are now required as stable releases (3.x and 2.x) rather than the previous pre-release versions.

Fixed

  • Re-running the Defect Explainer on the same test case replaces the previous explanation instead of appending a second copy below it.
  • Explanations containing < or & no longer break the layout of the execution comment they are written into.
  • An execution comment that is empty is left untouched instead of being replaced by a lone result heading.
  • Failure traces passed to the LLM keep table rows on one line and no longer contain non-breaking spaces, so the model sees the trace as readable plain text.

1.0.1 - 2026-06-29

Fixed

  • Use the current version of a test case set instead of the checked-in version.
  • Do not automatically switch to a checked-in version just because one exists.

1.0.0 - 2026-06-12

Added

  • Initial public release as open source project
  • Three AI agents: test case set reviewer, test case set describer, and defect explainer
  • Pluggable LLM provider architecture with built-in support for OpenAI, Azure OpenAI, and Anthropic
  • YAML-based prompt templates with Jinja2 support for full customization
  • Per-project overrides for language and LLM configuration
  • German and English locale support
  • REST API with built-in Swagger UI at /docs
  • SSL/TLS support with optional mutual TLS (mTLS)
  • Trusted reverse proxy configuration
  • CLI commands: init (scaffold config) and start (run the service)