I built something I needed: a UX writing agent for Claude that reviews interface copy directly from Figma.

Here’s what it does — you drop in a Figma frame link, and Claude uses the skill to audit your copy against UX writing best practices. Error messages that lack actionability. CTAs that don’t set clear expectations. Microcopy that creates cognitive load instead of reducing it.

This matters because we’re at an inflection point. AI tools are everywhere, but most content designers are still operating like it’s 2019 — manually reviewing every screen, writing similar strings over and over, burning cycles on work that should be systematized.

I’m releasing this as open source because I think our discipline needs to own how AI changes our work. Not defensively, not by pretending LLMs will replace judgment, but by building the tools that let us operate at a different level — less time on mechanical review, more time on strategic problems that require human expertise.

The technical foundation is Claude’s MCP + skills system. The content framework draws from years of practiced critique and established UX writing principles. The result is something that catches 80% of the obvious issues so you can focus on the 20% that actually requires your expertise.

If you work in content design, UX writing, or any adjacent discipline, the code is on GitHub. Contributions welcome — this gets better the more experienced practitioners feed into it.

🌐 Website: content-designer.github.io/ux-writin…

🐯 GitHub repo: github.com/content-d…

This is how I think content design evolves: not by resisting AI, but by building systems that amplify human judgment instead of replacing it.

Agree?