False affordance and anti-affordance
SkillDev toolsUse this skill when designing or fixing controls that mislead users, elements that look interactive but aren''t (false affordance), or controls that need to clearly communicate "you can''t do that here" (anti-affordance, like disabled states). Trigger when reviewing UIs where users tap inert elements expecting response, when designing disabled states, when picking treatments for read-only fields, or when a "decorative" element is being mistaken for a button. Sub-aspect of `affordance`; read that first.
Use False affordance and anti-affordance in Claude, ChatGPT or Ahel Desktop
Free. Sign in, add False affordance and anti-affordance and connect your AI. About a minute.
Also: Claude Code · Cursor · Codex
Then ask your AI: use the False affordance and anti-affordance skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; Ahel provides instructions and does not run this skill.
No other account needed.
Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
What this skill tells your AI
The instructions your AI receives, as published by hashgraph-online/awesome-codex-plugins in plugins/HDeibler/universal-design-principles/plugins/interaction-and-control-principles/skills/affordance-false-and-anti/SKILL.md and read by Ahel’s review.
The mirror images of standard affordance: the elements that look interactive but aren't, and the elements that are interactive but should signal "not right now." Both deserve as much design attention as standard affordance — and both are routinely neglected.
False affordance: looks interactive, isn't
A false affordance is an element whose visual treatment suggests interactivity it doesn't have. The user is led to expect an action; the action doesn't happen; trust drops.
Common sources:
- Cards and panels styled like buttons. A bordered, rounded, slightly-elevated container holds static content. Users tap; nothing happens.
- Hover states on non-interactive elements. A box that highlights on hover but isn't clickable. Users learn to mistrust hover signals.
- Hyperlink-blue text that isn't a link. Some design systems use blue for emphasis without realizing users read it as link affordance.
- Underlined non-link text. Underline is a strong link signifier; using it for emphasis or definition collides with that convention.
- Icon-buttons next to non-interactive content. A trash icon shown next to an item the user actually can't delete.
Fixes
- Strip interactive-looking treatments from inert content. Use neutral backgrounds, no borders, no hover changes.
- Reserve underlines for actual links. Use bold or italics for emphasis instead.
- If a box must look distinct, use uniform connectedness (a tinted region) rather than button-like elevation (drop shadow, border).
Anti-affordance: signaling "you can't"
Anti-affordance is the deliberate design of cues that tell the user "this is interactive in principle but unavailable to you right now." The most common case: disabled controls.
A disabled control needs to:
- Look different from enabled (lower opacity, neutral color, no fill).
- Not respond to hover or click as it would when enabled.
- Show a "no" cursor (
cursor: not-allowed). - Optionally explain why it's disabled — via tooltip or adjacent text.
<button class="btn" disabled
aria-disabled="true"
title="Add at least one item to checkout">
Continue to checkout
</button>
.btn[disabled] {
background: hsl(0 0% 92%);
color: hsl(0 0% 50%);
cursor: not-allowed;
opacity: 0.7;
}
.btn[disabled]:hover {
/* no change — this is the anti-affordance */
background: hsl(0 0% 92%);
}
When not to disable
A common UX mistake: disabling a button silently when the user "isn't ready" (form has unfilled required field; cart is empty). The user sees a non-responsive button and doesn't know why.
Better:
- Keep the button enabled and surface the validation error on click. The user gets a clear explanation rather than a silent block.
- Or, if the button must be disabled, explain why alongside it (a small note, a tooltip on hover/focus, or a visible reason).
<button class="btn">Continue to checkout</button>
<p class="hint">Add at least one item to your cart to continue.</p>
The user can read the hint and act; the button isn't a dead end.
Read-only fields
A field that's input-shaped but not editable is a special anti-affordance case.
<label>Account ID
<input type="text" value="acct_8f3c7a" readonly />
</label>
input[readonly] {
background: hsl(0 0% 96%);
color: hsl(0 0% 30%);
cursor: text; /* still selectable for copy */
border: 1px solid hsl(0 0% 90%);
}
The treatment says "this is data, not a control" — neutral background, no focus outline activation, but still selectable so users can copy.
For fields that aren't currently editable but could be (e.g., locked behind a permission), the visual treatment should hint that editability exists in principle.
False affordance from animation
A subtle modern source: animating a non-interactive element. A heading that wiggles, an icon that pulses, a card that slowly tilts. These attract attention; users assume attention means interactivity.
If the animation is decoration only, audit whether it's drawing attention away from real interactive controls. If it's a callout for new content, ensure clicking does something — even if just to dismiss the callout.
Worked examples
Example 1: a card that should be clickable vs. one that shouldn't
<!-- Clickable: card-link wraps, hover responds -->
<a href="/projects/acme" class="card-link">
<article class="card">...</article>
</a>
<!-- Inert: no link wrap, no hover response -->
<article class="card card--inert">
<h3>System status</h3>
<p>All services operational.</p>
</article>
.card { padding: 16px; border: 1px solid hsl(0 0% 88%); border-radius: 8px; }
.card-link { display: block; text-decoration: none; color: inherit; }
.card-link:hover .card { border-color: hsl(220 90% 60%); box-shadow: 0 4px 12px rgba(0,0,0,0.08); }
.card--inert {
background: hsl(0 0% 98%); /* slightly different bg signals "informational" */
border-color: hsl(0 0% 92%);
}
.card--inert:hover {
/* no change — anti-affordance */
}
The clickable card shifts on hover; the inert one stays still. Users learn quickly that hover-responsive cards are clickable and others aren't.
Example 2: read-only vs. disabled vs. unavailable
Three subtly different states, each with its own anti-affordance:
<!-- Read-only: data, never editable here -->
<label>Workspace ID
<input value="ws_abc123" readonly />
</label>
<!-- Disabled because of state: editable when condition is met -->
<label>Email
<input value="user@example.com" disabled aria-describedby="email-locked" />
</label>
<p id="email-locked" class="hint">Verify your email to enable changes.</p>
<!-- Unavailable due to permission: visible but not actionable -->
<label class="field-locked">
<span>Workspace name <LockIcon aria-hidden /></span>
<input value="Acme Inc" readonly />
<p class="hint">Only workspace admins can change this.</p>
</label>
The user gets specific anti-affordance signals for each: read-only data is visually demoted; conditionally-disabled is enabled-looking-but-with-explanation; permission-locked has a lock icon and explanation.
Anti-patterns
- Decorative elevation on content cards. Drop shadows that suggest "object you can interact with" on cards that are pure content. Strip the shadow.
- Disabled-button silent dead ends. A submit button that's just disabled with no explanation. Users hunt for what's wrong.
- Read-only fields styled identically to editable. Users tab into them, type, get confused.
- Inert hover responses. A non-clickable element with a hover style applied "for visual interest." Users tap it, nothing happens, trust drops.
Heuristics
- The "I tapped, nothing happened" audit. Watch users. Each unproductive tap is a false-affordance failure or an anti-affordance failure (disabled with no explanation).
- The hover-style audit. Every element with a hover state should be interactive. If you have hover styles on non-interactive elements, strip them.
- The disabled-with-context check. Every disabled control on every page should answer "why?" — via tooltip, adjacent text, or context.
Related sub-skills
affordance(parent).affordance-signifiers— the standard case of making affordance visible.feedback-loop— disabled controls should still give some feedback (cursor change at minimum).accessibility-understandable(process) — disabled states must be programmatically conveyed (disabledattribute,aria-disabled).
Signals
- GitHub stars
- 1k
- Forks
- 316
- Last commit
- Oct 2026
Advanced
- Item type
- skill
- Key
affordance-false-and-anti- Source
- github.com/hashgraph-online/awesome-codex-plugins
github.com/hashgraph-online/awesome-codex-plugins