Files
windmill/ai_evals/cases
Ruben FiszelandClaude Opus 4.8 64c8f2beca test(ai-evals): restore focused welcome_email flow case
Keep a retained global case for quote-heavy inline flow creation, per the
ai-chat skill's requirement that a behavior change ship with an adjusted
eval case. Benchmarked before (main) vs after (this PR's head) on Sonnet
and GPT-4o: capable models inline unchanged (no regression) and the
softened guidance no longer forces the multi-round-trip fill path.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-22 14:23:47 +00:00
..
2026-04-30 12:23:59 +00:00