test-poc

Compare Hebrew quote extraction methods and generate a comparison report.

1|Updated Mar 25, 2026
One-click install
npx skills add https://github.com/unfoldingWord/bp-assistant-skills --skill test-poc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: test-poc
Source: https://github.com/unfoldingWord/bp-assistant-skills/tree/main/.claude/skills/test-poc
Command: npx skills add https://github.com/unfoldingWord/bp-assistant-skills --skill test-poc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill provides an automated AB test workflow to compare prompt-over-code versus script-based quote extraction methods in translation notes, enabling evidence-based improvement decisions.

Core Features & Use Cases

  • Enables side-by-side evaluation of Hebrew quote extraction approaches using PSA chapters.
  • Generates a consolidated report and optional AT-fit validation outputs for project QA.
  • Supports resolution of test data paths and re-usable test scenarios.

Quick Start

Run the test-poc to compare Hebrew quote extraction methods on a PSA chapter.

Frequently Asked Questions about test-poc

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I A/B test Hebrew quote extraction methods for translation notes?

To A/B test Hebrew quote extraction for translation notes, you can run an automated workflow that compares a legacy script against a semantic alignment approach on PSA chapters. It generates per-verse quotes, a comparison report, and optional AT-fit validation outputs.

What's the best way to compare prompt-based versus script-based quote extraction?

Comparing prompt-based versus script-based quote extraction is best handled by an automated A/B test workflow that evaluates both methods side-by-side on the same PSA chapter data. This produces a consolidated report highlighting differences for translation notes QA.

Can I validate AT-fit for translation notes alongside quote extraction?

Yes, you can validate AT-fit for translation notes alongside quote extraction when prepared notes are present. The A/B test workflow outputs optional AT-fit validation results in addition to the standard per-verse quote comparison report.

Does the Hebrew quote extraction test require specific input data paths?

Hebrew quote extraction testing requires resolved data paths and validated required inputs for PSA chapters. The workflow automatically resolves test data paths and validates inputs before writing the structured report to the output directory.

How does semantic alignment compare to legacy scripts for extracting Hebrew quotes?

Semantic alignment compares to legacy scripts by evaluating both methods against PSA chapters to produce per-verse quotes. The A/B test generates a structured comparison report so you can see exactly how the semantic approach performs against the legacy script.

Where does the A/B test report for Hebrew quote extraction get saved?

The A/B test report for Hebrew quote extraction is written as a structured markdown file to the /tmp/claude/poc_ab_report.md path. This consolidated report contains the per-verse quote comparisons and optional AT-fit validation results.