feat: add multi-turn conversation eval example with details field#507

Open

christso wants to merge 13 commits intomainfrom

feature/multi-turn-conversation-eval

Commits on Mar 9, 2026

feat: add details field to freeform llm-judge schema

christso
and
claude
committed
docs: add multi-turn conversation eval dataset with composable asserts
christso
committed
docs: add README for multi-turn conversation evaluation example
christso
committed
docs: add llm-judge prompt templates for multi-turn conversation eval

christso
and
claude
committed
style: fix biome formatting for details field chain

christso
and
claude
committed
test: add e2e baseline and fix judge templates to include {{ answer }}

christso
and
claude
committed
fix: pass full conversation history to judge templates via {{ input }}

christso
and
claude
committed
chore: remove auto-generated baseline yaml (jsonl is canonical)

christso
and
claude
committed
fix: evaluate all assistant turns across full conversation

christso
and
claude
committed
feat: add {{ conversation }} template variable for role-annotated input

christso
and
claude
committed
feat: include role annotations in {{ input }} template variable

christso
and
claude
committed
fix: resolve file references in {{ input }} judge template variable

christso
and
claude
committed
fix: use string literal instead of template literal in test

christso
and
claude
committed