How this rule is fixed

This is an AI-enabled rule. Pass/fail requires model judgment; there is no deterministic fixer in the pilot registry.

• Harness: invoke-ai-ruleset-harness.sh runs design-rules/ai/run-design-ai-rule.sh on the Before fixture and expects findings with matching principleId.
• Remediation: Cursor agent plans from forge-ux-remediation.plan.md after sitewide audit — not handbook After copy.

Detection module: docs/design/ux-audit/ai-enabled-design-principles.md#ai-context-burden-subjective. Scroll down for Before / After examples and Evidence and remediation steps.

Purpose

Kitchen Sink landing_page and product_page shells can satisfy DET.CONTEXT.BURDEN numeric caps—hero interactive controls ≤ 3, header nav links ≤ 7, pre-main link clusters within budget—while still feeling overwhelming on screenshot review. This AI rule catches subjective overload that counts miss: stacked hero copy bands, dense diagrams beside headlines, multiple equal-weight cards in the first viewport, and ambient or decorative layers that compete with the story.

Deterministic burden checks enforce enterprise first-screen budgets; this rule enforces perceived load. A page may pass every threshold yet fail if a first-time reader cannot pick a single focal path before the fold.

Plan: Capture the first viewport and ask whether one story wins—or whether hero copy, diagram, cards, and chrome all shout at equal volume. Do: Stage the product story (hero → outcomes → mechanism → depth); defer dense diagrams and link clusters below the fold. Check: Re-read with fresh eyes; confirm DET.CONTEXT.BURDEN still passes after edits. Adjust: If the same overload pattern repeats (for example hero copy band count or diagram-in-hero), propose a deterministic DET.* companion.

Passing signals

• First viewport has one dominant focal path: gradient product-landing-title → short landing-hero-tagline → primary CTA pair—not six parallel copy bands.
• Hero interactive controls stay within budget and read as a clear next step (btn btn-forge + one secondary), not a button row plus a secondary link wall.
• Product visuals (landing-hero-visual, landing-forge-visual from render_landing_signal_field) support the headline; dense forge-diagram / ks-diagram-tile topology is deferred to a labeled section below the fold.
• Outcome or proof content lives in the next forge-section with section-label, not three equal-weight forge-card tiles competing with the hero in the same viewport.
• Ambient layers use restrained modifiers (forge-ambient--subtle) when explanatory copy is present; expressive mesh stays out of text-heavy heroes.
• Each visible band has a single job (DET.SECTION.SINGLE_JOB); navigation chrome is curated, not duplicated as inline link clusters inside the hero.

Failing signals

• Hero stacks kicker, two-line title, tagline, clarification, explainer, audience line, CTA row, secondary links, and support bullets—reads as a wall even with ≤ 3 buttons.
• Large labeled architecture diagram (ks-diagram-tile, forge-diagram) sits beside the headline at hero scale, forcing split attention before the promise lands.
• Three or more forge-card outcome tiles with card-amber accents share the first viewport with the hero at equal visual weight.
• Competing section-label headings, announcement strips, and hero kickers create multiple entry points with no clear read order.
• Numeric caps pass (DET.CONTEXT.BURDEN green) but screenshot review still feels like documentation cover or link-wall lite—visual noise and competing focal points dominate.

Before example

Before (failing example)

Forge Platform

Governed delivery spine for human-owned agent execution

Intent, structure, delegation, review, and evidence in one workspace.

Not a docs tree—a product control plane you can inspect.

Lenses shows workspace state; LCDL governs LLM tasks; Fleet runs bounded jobs.

For platform engineers, operators, and methodology leads.

Start quickstart Read docs

Trust · Schemas · Operate

• Local-first workspace visibility
• Governed LLM task layer
• Token-protected job plane

Methodology

ForgeSDLC

Intent to evidence

Control plane

Lenses

Workspace visibility

Reasoning

LCDL

Governed LLM tasks

After example

After (passing example)

Forge Platform

Governed delivery spine

Human-owned intent, agent execution, and reviewable evidence—without opening the full handbook first.

Start quickstart Read docs

One governed-flow snapshot—not the full architecture map.

Outcomes

Three layers, one story

Methodology

ForgeSDLC

Shape intent and release with evidence

Control plane

Lenses

See workspace state and next safe action

Reasoning

LCDL

Versioned, schema-aware LLM tasks

How it works

Platform topology

Full subsystem map for readers who already understand the promise.

Evidence and remediation

Capture: desktop first-viewport screenshot; note whether DET.CONTEXT.BURDEN passed. Crop hero copy column, hero visual, and any cards visible without scrolling. Record competing focal points and copy band count.

Remediate (in order):

1. Collapse hero copy to kicker + headline + one tagline; move clarification, ecosystem lists, and support bullets below the fold or into outcome cards.
2. Replace hero-scale dense diagrams with a signal visual (landing-forge-visual / product screenshot); defer full ks-diagram-tile to a labeled How it works section with intro copy.
3. Move outcome forge-card rows into the next forge-section; keep ≤ 3 cards per row and one section-label per band.
4. Step ambient intensity down (forge-ambient--subtle) when hero copy is long; add scrim/card surfaces if mesh remains visible (AI.AMBIENT.READABILITY_CONFLICT).
5. Re-run DET.CONTEXT.BURDEN; if subjective overload repeats with stable DOM signals (hero copy band count, diagram-in-hero), propose a deterministic companion (for example max hero copy paragraphs above CTA).

Related rules

• DET.CONTEXT.BURDEN — numeric first-screen / chrome caps (hero controls, nav links, link clusters).
• AI.CONTEXT.COGNITIVE_CLARITY — first-time readers can form a correct mental model in one pass.
• AI.NARRATIVE.COHERENCE — sections follow problem → outcome → mechanism → next step.
• AI.VISUAL.HIERARCHY_CONFIDENCE — scale, contrast, and grouping establish an obvious focal path.
• DET.SECTION.SINGLE_JOB — each section band serves one purpose, not stacked competing jobs.