The Benchmark Runner
ZIPThe automated version as a folder you unzip. Paste a key, double-click, and score in your browser. For people comfortable pasting a key into a file.
Downloads
All free. The lab and the prompts are also on this site, ready to copy.
The automated version as a folder you unzip. Paste a key, double-click, and score in your browser. For people comfortable pasting a key into a file.
For the by-hand route. It relabels a folder of answers as A, B, C and keeps the key until you ask for it. Works with text, images, and web pages.
The whole method as a skill file for an AI agent tool such as Claude Code, Cursor, or Codex. It interviews you, builds your task set, and guides the comparison. In a chat tool, paste the file and say "act as this skill".
The full lab with all seven prompts, for printing or reading offline.
The session in article form.
The method on one page.
Keep going with us
Cohort 6 · starts October 5
Go deep on building and leading agents, and design your organization's agentic operating system.
Sign upCohort 4 · starts October 12
Need to catch up first? Get fluent in using AI and building with AI, for yourself.
Sign up