PR Evidence Builder

📋 PLAN.md

Incomplete

Task Description

Primary Metric

Target Value

Base Checkpoint

Strategy

Why?

Evaluation Slices (comma-separated)

Incomplete

Decision Title

Context (Why is this decision needed?)

Alternatives Considered

Decision & Rationale

Consequences

Incomplete

Slice	Metric	Value	Notes

Incomplete

Document 10 failure examples with labels and analysis.

Incomplete

80-120 words. Be specific and honest about what doesn't work.

0 words (target: 80-120)

Incomplete

Environment Setup

Training Command

Evaluation Command

Seed

Python Version

Preview

Select a tab to preview the generated markdown.

## Summary

What changed and why (1-2 sentences)

## Evidence

Link to METRICS.md table

## Error Analysis

Link to FAILURES.md with themes

## Limitations

Link to LIMITATIONS.md

## Repro Steps

Link to REPRO.md

## ADR Link

Link to decision document