v0.0.240
·
1810 commits
to main
since this release
Version 0.0.240 (November 14, 2025)
- Introducing Evals! You can now create and run custom sets of tests on demand to assess system performance. Define your own test cases and assertions, such as:
- User prompts triggering create_data on table1 and table2
- Validating that specific data columns (e.g., a, b, c) are present
- Using custom LLM Judge prompts to automatically determine pass/fail outcomes
- Added the ability to adjust the sample k size for schema tables and metadata resources
- Improved the data source pages for a faster, smoother experience, including enhanced loading indicators and improved item removal
- Unused steps are now auto-deleted after 14 days. You can restore them anytime by rerunning the code.