Module reflection_eval 

Fixture-based reflection evaluation harness.

This module lets us score the current reflection heuristics against canned scenarios without requiring a live model/provider call.

Structs

ReflectionEvalCase

A canned scenario for evaluating reflection-guided retry quality.

ReflectionEvalReport

Report for one reflection evaluation case.

ReflectionEvalSummary

Summary report for a batch of reflection evaluation cases.

ReflectionEvalToolResult

Compact fixture representing one tool result for an evaluation turn.

ReflectionEvalTurn

Compact fixture representing an initial or revised answer.

Enums

ReflectionEvalToolOutcome

Tool outcome used in a fixture-based reflection evaluation turn.

Functions

builtin_reflection_eval_cases

Built-in scenarios for regression-checking reflection effectiveness.

evaluate_reflection_case

Evaluate one canned reflection case.

evaluate_reflection_cases

Evaluate a batch of reflection cases.