Craft Guard logo

Craft Guard

Community
Bandersnatch0x
craft-guard

UI craft and feedback quality. Use when polishing product UI (motion, loading and error feedback) or when a page reads as AI slop.

Overview

PublisherBandersnatch0x
Repositorydesign-playbook
Skill namecraft-guard
Stars
3
Forks
0
Bundled files
2
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.

  • 2 bundled files

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

  • Open source

    Published by Bandersnatch0x on GitHub. Read the source before you install it.

Installation

Install the Craft Guard 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/Bandersnatch0x/design-playbook/tree/main/packages/design-playbook/skills/craft-guard
  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/Bandersnatch0x/design-playbook.git /tmp/design-playbook
mkdir -p .claude/skills
cp -r /tmp/design-playbook/packages/design-playbook/skills/craft-guard .claude/skills/craft-guard
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Craft Guard 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 Craft Guard 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 Craft Guard 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.

craft-guard

Craft declaration for cross-product quality: hierarchy, wait feedback, purposeful motion, CJK-safe type. Does not own business risk (domain) or token inventory (design).

When a verified design-baseline binding exists (status: ready), use that path as the project-specific visual baseline. Report clear divergence (density, type, spacing, shape, motion, or shared component treatment) to ui-evaluator with source set to the bound path. Do not replace valid project choices with this skill's generic taste defaults.

When a persistent contract v1 is bound for the run (contract-bind.json / scripts/contract_v1.py), honor decided visual and craft constraints from that contract and reject unknown contract schema versions. Do not invent page or component inheritance layers.

Apply (checklist)

Treat as exhaustive for the surface under edit.

Hierarchy and type

  • One clear primary. Secondary recedes via density and space, not rainbow chrome.
  • CJK UI uses the project Chinese stack (var(--font-cn) or equivalent).
  • Brand solid + brand-gradient ≤ 3 emphasis hits per viewport. Neutrals carry the rest.

Done when: a cold reader can name the primary action and region in one glance.

Loading tiers

Tiers and their boundaries are the threshold of PERF-01@2 in ../design-playbook/references/rules.md, the single source for these numbers:

text
below placeholder-from           no indicator
placeholder-from to indicator-from   skeleton (layout stable)
past indicator-from              loader + what is happening
past timeout-at                  timeout + retry (or cancel)

Done when: every async path in scope maps to a tier with a next action on failure/timeout.

Motion

  • UI motion stays within motion-max, and hover, focus and typing feedback within high-frequency-motion-max (the threshold of CRAFT-08@2). Animate transform / opacity only.
  • Each animation names the state change it explains.
  • Honor prefers-reduced-motion. Keyboard-triggered actions stay free of decorative motion.

Done when: every motion in scope states the state change it explains.

Replace AI slop with the target look

For implemented UI, evaluate registry-entry applicability predicates in ../design-playbook/references/rules.md. Use the subset declared by the active spec and contract, or the full catalog when none is declared. Run every entry whose predicate evaluates to applicable against rendered UI plus relevant source. Execution protocol and the row format live in references/detectors.md. Write exactly one seven-column audit row per applicable entry to .scratch/<run>/craft-guard.md:

ID@verApplicabilityPredicate reason / missing proofResultRendered evidenceSource evidenceException checkPositive fix

Applicability is the entry's three-state predicate outcome: applicable, not-applicable, or blocked. not-applicable and blocked both require an observable reason (blank is invalid, never a silent skip). Result is clear|hit only for applicable rows, - otherwise; Positive fix is required on hit rows. Missing rendered or source proof is blocked, not a silent clear. Unknown registry IDs fail at this stage. Audit rows are advisory: record evidence, exception check, and positive fix. Leave declaration source, severity, and verdict to ui-evaluator.

The host model may have no vision (text-only input). Renders stay path-bound: the Rendered evidence column cites the artifact path; never read the image. Assertions run on the text face: HTML and CSS source, a11y tree text, and interaction-trace action records. An interaction trace is a Playwright trace ZIP. Unzip before reading; the artifact is not JSON. A no-vision run evaluates the same registry protocol this way without degrading it. Genuinely missing rendered proof still records blocked honestly.

Push towardInstead of default sludge
Accent on key noun + primary CTAPurple–blue gradient wallpaper
Glass only on rare floating layersBlur on every card
Project CJK stackInter and Roboto as Chinese body
Weighted modulesEqual white card grid
State-explaining motionBounce and elastic decoration

references/craft.md, required when the surface has L4 interactive zones (per-zone hover and motion affordance, with its own Done when). Also holds edge craft (failure and permission feedback, rounded-corners and shadows, charts).

Completion

Done when: hierarchy, every in-scope wait and fail path, and every animation pass their checks. If the surface has L4 interactive zones, every zone's affordance passes its checks (per references/craft.md). Implemented UI also records exactly one seven-column audit row for every registry craft entry whose applicability predicate evaluates to applicable, with no unexplained missing proof. Residual hits and blocked proof are handed to ui-evaluator. Audit rows do not assign source, severity, or verdict. Machine check right after writing craft-guard.md: python scripts/g8_run_registry.py .scratch/<run>/craft-guard.md.

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 Craft Guard AI skill do?

UI craft and feedback quality. Use when polishing product UI (motion, loading and error feedback) or when a page reads as AI slop.

Why use Craft Guard on TypingMind?

Because you install it once and use it with any model. Craft Guard 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 Craft Guard in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Bandersnatch0x/design-playbook/tree/main/packages/design-playbook/skills/craft-guard. 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 Craft Guard?

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 Craft Guard?

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

Is the Craft Guard AI skill free?

It is published on GitHub by Bandersnatch0x. Check the repository for licensing terms. 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 👇