wtf is gogh?

Gogh is a tool that combines six separate frontend design skill packages into one organized reference system, stored as an Obsidian notes vault, meant to give AI coding agents better design judgment. Instead of relying on a single prompt to make a website or app look good, Gogh tells an AI agent which of the six design skills to use for a given task, how to combine them, and how to resolve cases where they disagree with each other. The six underlying skills each come from a different author and cover a different part of frontend design work: setting an overall aesthetic direction, executing small interactive details like button press states, enforcing consistent design rules across a codebase, looking up styles and font pairings, and auditing a finished interface against design guidelines. Gogh keeps track of the license and source of each skill and records when two skills give conflicting advice rather than hiding the disagreement. Gogh is a command line Python tool. You install it with pip and can try a demo, lint an example project, or generate a report against a sample vault, all built in. To use it on a real project, you create a new vault for a client, feed it a source document, then run its synthesis, visuals, and report commands to produce stack advice and a checklist tailored to that project. The project is explicit that its current version is read-only and advisory: it does not automatically change your code, accounts, or any live production systems, it only produces recommendations and reports for a human or an AI coding agent to act on. It defines a five-stage maturity system, from an early scaffold up to fully market-ready, and a release script checks for accidentally included secrets or local file paths before packaging a release. Gogh is released under the Apache 2.0 license, a permissive license that allows free use including commercial use as long as you keep the required notices. It is aimed at developers, indie hackers, design-focused engineers, and small agencies who build frontends with AI coding tools such as Claude Code, Cursor, or similar assistants, and who want the results to look deliberately designed rather than generic.