Refine logo

Refine

Community
Darcos-Loft
refine

A shared vocabulary of named refine commands for existing UI (typeset, colorize, animate, settle, flow, bolder, quieter, distill, regroup, glass, iconify, deslop, brandward, productward, audit, critique, harden, polish). Use when the user asks to improve, fix, tighten, calm, strengthen, or polish a UI with a precise verb instead of a vague "make it better". Each command is one scoped move wired to the rest of the suite.

Overview

PublisherDarcos-Loft
Repositoryfluid
Skill namerefine
Stars
3
Forks
0
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Darcos-Loft on GitHub. Read the source before you install it.

Installation

Install the Refine AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Darcos-Loft/fluid/tree/main/refine
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Darcos-Loft/fluid.git /tmp/fluid
mkdir -p .claude/skills
cp -r /tmp/fluid/refine .claude/skills/refine
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Refine in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Refine on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Refine is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Refine

A shared vocabulary of small, named moves you make on UI that already exists. Each command is one verb with one job. "Make it better" is vague and drifts; typeset then quieter then polish is precise, repeatable, and reviewable.

This is the granular layer of the suite. design-system creates an identity, fluid sets the motion, output gates the ship. refine is the set of verbs you use in between to push a screen from done to right.

How to invoke

The user names the verb, in their words or as a slash command:

  • "typeset this", "make the hero bolder", "calm the motion", "/polish the pricing section"

Apply exactly that move, nothing else. Do not colorize when asked to typeset. If the user named a region (a section, a component, a selection), scope to it; otherwise apply to the current view. State the change in one line, then run audit if motion or color was touched.

Commands compose. Chain them in the order the user implies, or in the natural order: structure, then type, then color, then motion, then polish. Full definitions, with what each one enforces and leaves alone, are in reference/commands.md.

The vocabulary

Type

  • typeset — repair typography: a distinctive pairing from the database, a real scale, measure under ~75 characters, tracking eased on large headings. Kills justified, sub-10px, and crushed text.
  • recase — fix casing tics: drop all-caps body, keep uppercase kickers to one if at all.

Color

  • colorize — apply or repair the palette: commit to a dominant with one rationed accent, near-black not pure black, no purple-wash gradient, body contrast 4.5:1.
  • tone — shift the mood (warmer, cooler, darker, more muted) without breaking the locks.

Motion

  • animate — add motion on purpose, with the house easings and a reduced-motion path. Reveals, heroes, hovers, counters.
  • settle — calm janky or excessive motion: instant on frequent interactions, named transitions over transition: all, scale(0.93) not scale(0), fixed transform-origin.
  • flow — give many small animations one shared physics source (one scroll velocity, one progress value) so they read as one system.

Emphasis

  • bolder — increase force: widen the type jump, raise contrast, give one element a real moment.
  • quieter — dial it down: ration the accent, hairlines over fills, more whitespace, fewer effects.

Structure

  • distill — strip slop structure: flatten nested cards, vary identical card grids, drop the icon-tile-above-every-heading and the 01/02/03 markers. Cut copy to the essential.
  • regroup — fix hierarchy and sectioning: correct heading order, group what belongs together, separate brand from product.

Surface

  • glass — apply the liquid-glass treatment to signal depth or dismissal, not as decoration.
  • iconify — replace emoji and stock icons with a consistent bespoke SVG set.

Copy

  • deslop — kill marketing buzzwords, manufactured "not X but Y" contrast, and em or en dashes. Say the concrete thing.

Register

  • brandward — push toward impression-led brand work (a landing page): more air, more type, a stronger hero moment.
  • productward — push toward task-led product UI (a dashboard): denser, calmer, faster, information-first.

Gates

  • audit — run the deterministic detector (29 rules) and report.
  • critique — qualitative review against the by-hand catalog (the structural and semantic tells the detector cannot see).
  • harden — accessibility and robustness: reduced-motion in CSS and JS, visible focus, contrast, color-scheme.
  • polish — the final craft micro-pass: press and hover feedback, optical alignment, consistent radii and spacing, the details that separate a demo from a shipped build.

Composition

  • New screen lands generic: distill → typeset → colorize → polish.
  • Motion feels off: audit → settle → flow.
  • Hero is flat: bolder → animate → polish.
  • Landing page reads like a dashboard: brandward → typeset → glass.
  • Before shipping: audit → harden → polish, then the output gate.

Wiring

These commands are not new rules. They are named entry points into rules the suite already holds.

  • typeset and colorize draw from design-system/data/ (the palette and type-pairing database) and design-system/reference/rules.md.
  • animate, settle, and flow apply the fluid SKILL rules and the house easings.
  • glass and iconify use design-system/reference/glass.md and reference/icons.md.
  • audit runs fluid/detector/detect.mjs. critique reads fluid/reference/anti-patterns.md.
  • harden and polish enforce the output checklist.

Pairs with: every skill in the suite. refine is the verb layer over all of them.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Refine AI skill do?

A shared vocabulary of named refine commands for existing UI (typeset, colorize, animate, settle, flow, bolder, quieter, distill, regroup, glass, iconify, deslop, brandward, productward, audit, critique, harden, polish). Use when the user asks to improve, fix, tighten, calm, strengthen, or polish a UI with a precise verb instead of a vague "make it better". Each command is one scoped move wired to the rest of the suite.

Why use Refine on TypingMind?

Because you install it once and use it with any model. Refine is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Refine in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Darcos-Loft/fluid/tree/main/refine. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Refine?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Refine?

As many as you like. As long as a model supports skills, you can use Refine with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Refine AI skill free?

Yes. It is published on GitHub by Darcos-Loft under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇