From the article Building for Agents: The Strategy for Cross-Industry Skill Standards →

Free Diagnostic Instrument

The Skill File Audit

Your team's best AI workflows probably live in someone's head, a personal notes app, or a Slack thread three forks deep. That works fine while a person runs them by hand. It stops working the moment an agent has to find that knowledge and run it alone, with nobody in the room to ask.

Get the Free Audit 8 pages · pay what you want, starting at zero

What's Inside

A Number, and Where It Came From

This is a scoring instrument, not a how-to guide. It tells you how much of your team's working knowledge sits in a form that something other than its author could pick up and run. It does not fix anything and it does not tell you what to build.

The Skill File Audit cover and scoring grid pages

What's Inside

  • A scoring grid for up to ten AI workflows your team already runs
  • Four dimensions, scored 0 to 3: trigger routability, body executability, version history, named owner
  • Anchor descriptions for every score, so two people scoring the same workflow land on the same number
  • A totals table and four scoring bands that say what the average means
  • A cross-platform reference mapping the four dimensions to Claude, OpenAI, Google Gemini, and teams running no agent runtime at all

A prompt nobody else can find is a prompt your team owns exactly once.

4

Dimensions scored on the same 0 to 3 scale

10

Workflows you can score in one sitting

8 pages

Rubric, scoring grid, and platform reference, nothing else

Fit

Who This Is For

Score your workflows if

  • Your team already runs several AI-assisted workflows and nobody has checked whether an agent could find and run them without a person in the middle
  • You want a number you can rescore in a quarter and compare against, not a one-time opinion
  • You are deciding where to invest in documentation and want to know which of the four dimensions is actually dragging your score down
  • You work on Claude, OpenAI, Google Gemini, or a setup with no agent runtime at all, and want the same rubric to apply

Skip this if

  • You want a guide that tells you how to write better prompts or skill files. This audit diagnoses; it does not prescribe a fix
  • You are not running any AI-assisted workflows yet. There is nothing here to score until there is something to score
  • You already have a formal documentation review process that covers routability, executability, version history, and ownership. You have already built what this measures
You cannot hand an agent knowledge your team has never written down.
From the audit's closing note
Get the Free Audit