Loading States logo

Loading States

Community
sanky369
loading-states

Audit an app's async surfaces and implement the right loading, progress, and empty states — skeleton screens, spinners, progress bars, first-use/no-results/error-empty views — chosen by expected latency and layout knowledge. Use when the user says 'add loading states', 'the page just goes blank while fetching', 'add a skeleton screen', 'users think it's frozen', 'spinner or skeleton?', 'design the empty state', 'zero state', or when reviewing an app whose fetches show nothing. Produces a loading/empty-state audit of every async surface plus the implemented components.

Overview

Publishersanky369
Repositoryvibe-building-skills
Skill nameloading-states
Stars
31
Forks
7
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 sanky369 on GitHub. Read the source before you install it.

Installation

Install the Loading States 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/sanky369/vibe-building-skills/tree/main/skills/frontend-design/loading-states
  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/sanky369/vibe-building-skills.git /tmp/vibe-building-skills
mkdir -p .claude/skills
cp -r /tmp/vibe-building-skills/skills/frontend-design/loading-states .claude/skills/loading-states
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Loading States 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 Loading States 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 Loading States 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.

Loading & Empty States

Make sure no user ever stares at a blank or frozen-looking screen: inventory every place the app waits or can be empty, choose the correct state for each by expected latency, and implement it. The prime directive: the indicator is chosen by how long the wait is and whether the layout is known — not by taste. Deliver an audit table covering every async surface plus the implemented states, never a lone spinner component.

When to use / when not to

Use for: loading indicators, skeleton screens, progress bars, empty states (first-use, no-results, error-empty, success-empty), "app looks frozen" complaints.

Hand off instead when the real need is:

  • Making waits shorter or invisible (optimistic UI, preloading, caching) → skills/frontend-design/performance-optimization — always consider that first; the best loading state is none
  • Error messages and recovery flows beyond a simple retry-empty-state → skills/frontend-design/error-handling-recovery
  • Shimmer/spinner motion polish, timing tokens → skills/frontend-design/interaction-physics

Step 0 — Inspect the codebase, then ask only what's left

  1. Inventory async surfaces: grep for fetch(, axios, useQuery/useSWR/createResource, isLoading/isPending/loading flags, Suspense boundaries, and route-level data loaders. List every view/component that waits on data or mutation.
  2. For each, note what currently shows while waiting: nothing (blank), layout jump, spinner, skeleton, or a disabled control. "Nothing" and "layout jump" are the defects to fix first.
  3. Check for existing skeleton/spinner/empty-state components — reuse and extend them rather than duplicating.
  4. Estimate expected latency per surface from evidence: cached client state (~instant), same-region API reads (usually sub-second), search/aggregation endpoints (seconds), uploads/exports/report generation (long, often measurable). Look for existing timeout configs or telemetry. Where you can't estimate, say so and pick the 1s-class treatment as the safe default.
  5. Enumerate empty-capable views: every list, table, grid, search result, and inbox can be empty. Check which ones currently render a bare nothing.

Ask the user (one batch, only if not inferable): whether any operations are known to be long-running (uploads, exports, batch jobs), and whether there's a design language for illustrations/empty states. Otherwise state assumptions and proceed — never block an audit on questions the code answers.

Workflow

1. Choose each loading state by the latency ladder

Expected waitTreatment
< 100msNothing — an indicator that flashes in and out is worse than none
100ms – 1sSubtle inline cue: button spinner/opacity, cursor: wait; no layout change
1s – 10sSkeleton if the incoming layout is known; spinner if not
> 10sProgress bar — determinate with count/percent if measurable, indeterminate + explanatory text if not

Fork rules:

  • Skeleton vs spinner: known content structure (cards, lists, tables, profiles) → skeleton mirroring that exact layout. Unknown/variable structure, or a small area like a button → spinner. Never skeleton modals, toasts, or dropdowns — those must feel instant.
  • Mutations vs reads: button-level loading (spinner in the button, label hidden, width preserved) for submits; page/section-level states for reads. Don't block the whole page for a local mutation.
  • Flash prevention: if a wait might be short, delay the indicator ~150–200ms so fast responses show nothing; once shown, keep it ≥300–500ms so it doesn't blink.
  • Determinate beats indeterminate whenever total size/count is knowable (uploads: bytes; batches: items done).

2. Design each empty state by type

Every empty view gets one of four treatments — classify, then apply:

TypeTriggerMust include
First useUser has no data yetWhat this area is for + primary CTA to create the first item
No resultsSearch/filter matched nothingEcho of what was searched + escape hatch ("Clear filters")
Error-emptyLoad failedPlain-language failure + Retry action (hand complex recovery to skills/frontend-design/error-handling-recovery)
Success-emptyAll items done (inbox zero)Positive confirmation; no CTA needed

Rule: an empty state without an action is a dead end — only success-empty may omit the CTA.

3. Implement with the structural rules

  • Skeleton dimensions match the real content's dimensions — the swap must not shift layout (this is also a CLS fix). Reserve space; vary text-line widths.
  • Show loading state immediately on action (or after the flash-prevention delay) — never wait for the first byte.
  • Accessibility: loading regions get aria-busy="true" and a polite live region announcing "Loading"; skeletons are aria-hidden="true"; progress bars use role="progressbar" with aria-valuenow/min/max; shimmer stops under prefers-reduced-motion.
  • Canonical implementations (skeleton shimmer, button spinner, determinate/indeterminate progress, all four empty states, a11y blocks): references/patterns.md — read it when writing the code.

4. Verify

Simulate slow network (devtools throttling or an artificial delay) and confirm: no blank frames, no layout shift at content swap, indicators don't flash on fast responses, empty states render with their actions.

Required output format

Deliver both artifacts:

1. The code — implemented skeleton/spinner/progress/empty-state components and their wiring into each async surface, reusing existing project components and tokens.

2. Loading & Empty State Audit (markdown):

## Async surface inventory
| Surface | Wait class | Current | Correct state | Status |
| Product grid | 1–10s (search API) | blank screen | card skeleton ×8 | implemented |
| Save button | 100ms–1s | none | inline button spinner | implemented |
| CSV export | >10s | spinner | determinate progress + count | implemented |
| Comments | unknown → assume 1s class | layout jump | skeleton, space reserved | implemented |

## Empty state coverage
| View | First use | No results | Error-empty | Success-empty |
| Projects list | ✓ CTA "Create project" | n/a | ✓ Retry | n/a |
| Search page | n/a | ✓ "Clear filters" | ✓ Retry | n/a |

## Assumptions
[Latency estimates you could not verify, flagged for the user]

Quality bar (check before delivering)

  • Every async surface in the inventory has an assigned state — no blanks left
  • Every state choice is justified by the latency ladder, not preference
  • Skeletons mirror real content layout; content swap causes zero layout shift
  • No skeleton on modals/toasts/dropdowns; no full-page block for local mutations
  • Indicators are flash-proofed (delayed show, minimum display time) where waits may be short
  • Every empty state except success-empty has an action
  • aria-busy, live-region announcement, role="progressbar" values, aria-hidden skeletons, reduced-motion handling all present
  • Operations > 10s show progress or explanatory text, never a bare spinner

Hard don'ts: no spinner-for-everything; no invented time estimates shown to users ("about 2 minutes") unless derived from real data; no loading state on operations that complete instantly.

Integration

  • skills/frontend-design/performance-optimization → run first or alongside; optimistic UI removes the need for many loading states, and skeleton space-reservation is shared CLS work.
  • skills/frontend-design/error-handling-recovery ← consumes your error-empty placements and expands them into full recovery flows.
  • skills/frontend-design/interaction-physics → supplies shimmer/spinner motion timing and the reduced-motion baseline.
  • skills/frontend-design/component-architecture → provides the component inventory your skeletons must mirror; add Skeleton/EmptyState as shared components there.

References

  • references/patterns.md — full implementation gallery: skeleton building blocks and card example, spinner and button-loading CSS, determinate/indeterminate progress bars, all four empty-state markups, and the accessibility blocks. Read it when writing the code.

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 Loading States AI skill do?

Audit an app's async surfaces and implement the right loading, progress, and empty states — skeleton screens, spinners, progress bars, first-use/no-results/error-empty views — chosen by expected latency and layout knowledge. Use when the user says 'add loading states', 'the page just goes blank while fetching', 'add a skeleton screen', 'users think it's frozen', 'spinner or skeleton?', 'design the empty state', 'zero state', or when reviewing an app whose fetches show nothing. Produces a loading/empty-state audit of every async surface plus the implemented components.

Why use Loading States on TypingMind?

Because you install it once and use it with any model. Loading States 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 Loading States in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/sanky369/vibe-building-skills/tree/main/skills/frontend-design/loading-states. 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 Loading States?

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 Loading States?

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

Is the Loading States AI skill free?

Yes. It is published on GitHub by sanky369 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 👇