Tastemaker is a workflow skill for coding agents that helps them turn brief-driven UI work into a consistent visual system with files, assets, contrast checks, and local project memory. It is aimed at developers using agents such as Claude Code, Windsurf, and Gemini CLI.
Tastemaker is a workflow skill for coding agents that helps them produce more consistent UI by grounding design choices in references, rules, and reusable project memory. The site positions it as a way to stop prompting for taste and instead give the agent a process it can execute.
In practice, Tastemaker reads the brief, defines the screens, locks the visual system, generates assets, builds the visual proof, and stores decisions for later use. The result is meant to live inside the project as files and assets rather than as a separate hosted service.
Tastemaker starts by scoping screens from the product brief before it generates color, type, and layout decisions. The page shows examples such as landing-page hero, pricing, CTA, onboarding states, dashboard states, and settings screens.
It can read reference pixels or generate a fresh palette, then compute legal color pairings against a contrast floor. The site explicitly shows WCAG-style contrast checks and a style-lock process.
The workflow sources a complete visual language, including icon families, imagery, a constructed mark, and favicons. The homepage says it avoids emoji placeholders and validated assets are included with the screen.
Instead of leaving claims as text, Tastemaker turns features into UI mockups, flows, charts, and before/after states. The homepage also notes that GSAP motion ships in the same pass and that reduced-motion behavior is considered on the comparison page.
Project decisions are saved locally so the agent can reuse preferences on future work. The site shows `.tastemaker/style-lock.md`, `decisions.log`, and a profile file that records keep/reject choices.
The product is designed to work with plain project files rather than a hosted black box. The pages show Markdown, Python, SVG, CSS, and JavaScript as the main output formats.
Use it when you want an agent to build a landing page or app screen with a more deliberate visual system than a default prompt usually produces. The site’s comparison page uses a coffee-subscription landing page to show the difference before and after the skill is installed.
Use it for multi-state product UI where empty, loading, populated, and settings views should feel like one system. The homepage explicitly lists onboarding, dashboard, and settings states as part of the scoped output.
Use it when you already have reference imagery and want the agent to derive a usable palette from the actual pixels. Tastemaker advertises reference reading, generated palettes, and contrast verification.
Use it for projects where the team wants the UI output stored as files in the repository, not locked in a hosted tool. The site shows local Markdown, logs, CSS, SVG, and JavaScript outputs beside the product code.
Use it when repeated UI tasks should carry forward previous choices instead of re-deciding them every time. Tastemaker saves keep/reject decisions into a local profile for future projects.
Tastemaker is installed as a skill for coding agents and triggers when you ask the agent to build, style, or improve a UI. The site shows a terminal-style install command using `npx skills add codeswithroh/tastemaker`, and notes that Claude Code can also use the plugin marketplace or a manual install.
The homepage describes Tastemaker as a workflow that grounds, checks, builds, and remembers design taste for a coding agent. Its steps include reading the product brief, locking the visual system, sourcing assets, building the proof in UI form, and saving decisions for future work.
The public pages show the output as local project files and assets, including `.tastemaker/style-lock.md`, `.tastemaker/decisions.log`, and generated design assets such as logos, icons, illustrations, CSS, SVG, Markdown, Python, and JavaScript.
The site says the free core stays free, while sponsor-exclusive aesthetic modes are available for $8/month. The pricing URL itself currently returns a page not found error, so there is no separate pricing page to review.
The homepage frames Tastemaker as agent-friendly for Claude Code, Windsurf, and Gemini CLI, and it also says the workflow triggers without a magic phrase, account, or hosted API. Beyond that, the site does not provide deeper integration documentation.
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.
SonOf connects to your repo and PM tool, audits your codebase, and turns approved work into shipped tickets with senior engineering review.
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.
