New aicoding-assistant

unslop

A coding-agent skill that turns generic, AI-looking UI into a designed interface.

Would you recommend unslop?

No votes yet. Tried it? Be the first to say.

One vote per person. No account needed.

Overview

unslop is a skill for Claude Code, Codex and Google Antigravity. It finds the tell-tale signs of vibecoded UI (purple gradients, glows, emoji labels, shouty copy, uneven spacing), sets a design direction from your product, then fixes the screen pass by pass like a designer's redline review. It works with Next.js, React, Vue, Svelte, Tailwind or plain HTML and CSS.

Instead of

Shipping the default look an AI coding agent produces, or a manual design review

Reach for it when

An AI agent built your dashboard or landing page and it looks like every other AI-built screen.

Key features

  • Finds the "slop"

    Spots the tell-tale signs of AI-generated UI, such as purple gradients, glows, emoji labels, shouty copy and uneven spacing.

  • Sets a direction

    Derives a design direction from your actual product rather than applying a generic theme.

  • Redline-style passes

    Fixes screens pass by pass, like a designer reviewing a build.

  • Framework-agnostic

    Works with Next.js, React, Vue, Svelte, Tailwind or plain HTML and CSS.

  • Works on live sites

    Can review a deployed page from a link when you don't have the code handy.

  • Multiple agents

    Packaged as a skill for Claude Code, Codex and Google Antigravity.

In depth

What it does

Coding agents are good at producing working interfaces quickly, but they converge on the same look: gradient hero sections, glowing cards, emoji in headings and marketing-style copy. unslop gives your agent a structured design review. It first identifies the specific patterns that make a screen look generic, decides on a direction that suits your product, then works through the UI in passes (layout, spacing, typography, colour and copy) rather than restyling everything at once.

Installing it in Claude Code

bash
# for all your projects
claude plugin install unslop --marketplace heydamir/unslop

# older Claude Code versions
claude plugin marketplace add heydamir/unslop
claude plugin install unslop@heydamir

To limit it to one project, run the install from that project's root with --scope local. To share it with your team, use --scope project and commit .claude/settings.json; each teammate then adds the marketplace once in their own copy.

Using it well

  • Point it at one screen or flow at a time and review the diff before moving on.
  • Give it context about your product and audience; the direction it picks depends on what it knows.
  • Keep your design tokens (colours, spacing, fonts) in one place so its fixes stay consistent across pages.

Strengths

  • Targets a very real problem: agent-built interfaces tend to look the same, and users notice.
  • Encodes design judgement as a reusable skill instead of a one-off prompt.
  • Works where you already are: inside your coding agent and on your existing codebase.
  • Easy to install per project, so a whole team gets the same design review.

Watch out for

  • Design is subjective; review the changes as you would a junior designer's suggestions.
  • Each pass spends agent tokens, so large apps cost more to "unslop".
  • It's a new and fast-moving project (released October 2026), so expect its behaviour to change.

Try it in a few minutes

Terminal
claude plugin install unslop --marketplace heydamir/unslop
# older Claude Code versions:
claude plugin marketplace add heydamir/unslop
claude plugin install unslop@heydamir

Alternatives

A design system (shadcn/ui, Radix, your own tokens)

Prevents generic UI by constraining choices up front.

A human design review

Still the gold standard for brand and product fit.

Other design-focused agent skills

Several coding agents ship or support front-end design guidance you can compare it with.

Related tools

Other tools that share topics with unslop.

Aider

aider.chat
Gem

AI pair programming in your terminal that edits files and commits to git.

Instead of Copying code between a chat window and your editor

llm (CLI)

llm.datasette.io
Gem

Use LLMs from the command line and pipe anything into them.

Instead of Copy-pasting terminal output into a chat window

promptfoo

promptfoo.dev
Gem

Test and compare LLM prompts and models with assertions, locally or in CI.

Instead of Eyeballing prompt changes in a playground