What is draft-review-kit?

Draft Review Kit is a collection of editorial review skills you can plug into AI writing environments like Claude Code or Codex. It is not a writing generator. Once you have a draft on the page, these skills act as a small panel of reviewers, each with a specific and distinct job, so you can run the kind of pressure you need rather than getting one generic round of AI feedback. The reviewers fall into four groups. Big-picture reviewers look at argument, structure, and stakes: one called dev-edit checks whether the piece holds together and pays off, another called guardrails scans for logic gaps, unsupported claims, and the kinds of patterned phrasing that tend to creep into AI-assisted writing. Two others, panel and debate, either synthesize feedback from multiple angles or have reviewers argue with each other until the tensions become clear. Reader-response reviewers simulate specific audience types. One reads as harshly as possible and attacks weak claims. One reads as a caring non-expert and flags anything that assumes too much. A third spots jargon, hand-waving, and steps the reader was expected to follow without explanation. Craft-lens reviewers channel the aesthetic priorities of particular writers or filmmakers: one checks for suspense and visible tension in the writing, one looks at pacing and forward motion, others look for humor, specificity, or unnecessary words. A final line-edit reviewer does a sentence-level pass and summarizes what it changed. The suggested workflow is sequential: structure first, reader-response testing second, craft lens third, a guardrails pass before publication, and a line edit last. Each skill is a plain Markdown file, which means it can be adapted to any AI environment that supports reusable instructions, not just the two it ships with. The project can also be forked and modified to match your own editorial standards.