Part of an experiment
This is the Oneshot build — a single Claude agent reads the plan and ships the whole thing end to end. See the video, blog post, and source code for the full study.
Blog post Video Source

CLAUDE.md Scorer

Score your CLAUDE.md and AGENTS.md files for quality, completeness, and effectiveness

📄

Drop your .md file here or click to upload

Supports CLAUDE.md, AGENTS.md, or any markdown instruction file

Or try a sample file:

Scoring your file...

📄 Scoring: -
-

Score Report

-

Findings

Suggestions for Improvement

Live Evaluation

🔒
Test how well agents follow your instructions with a live Claude evaluation
Your key is never stored -- used only for this session

Running task 1 of 3...