I made a Claude Code skill inspired by Jordan Peterson’s essay-writing guide

I made JP Essay, an open-source skill that takes Claude Code through research, outlining, drafting, revision, and fact-checking to produce one final essay.

The idea is to make the writing process inspectable. You can open the research notes, examine the outline, and follow the revisions that led to the final text.

Starting with a question, JP guides Claude through a sequence of writing tasks:

One part I find particularly interesting is the fresh-reader step. A separate agent reads the draft and reconstructs its argument from memory. That feedback helps identify ideas that were difficult to follow before another revision.

It’s an AI review, so it can still miss problems. The fact-checking stage also doesn’t remove the need to verify important claims yourself.

The implementation combines a SKILL.md file, intermediate Markdown artifacts, and a checking script that validates the required steps.

Keeping the intermediate work makes it possible to investigate where something went wrong. An unclear conclusion might come from the outline, a missing source, or a revision that weakened the argument.

JP produces one final essay per run. The comparison image shows adapted excerpts from two separate runs: one using direct writing and another using a short-format adaptation of the JP workflow.

JP is free and open source. You need your own Claude Code access; model usage isn’t included.

What would you test first to evaluate a writing workflow like this: factual accuracy, argument structure, readability, or the amount of editing it saves?

The inspiration came from Jordan Peterson’s essay-writing guide, particularly its emphasis on refining sentences and reorganizing paragraphs to clarify an argument. I wanted to bring that deliberate process into Claude Code. JP is an independent project with no affiliation with Jordan Peterson.