# Knowledge Graph Index

Karpathy-style compiled graph for this library. **Compile once, traverse forever** — no vector search required at this scale.

Generated: 2026-06-23T08:13:25.602Z

## How agents should query

1. Read this index for hubs and tensions.
2. Fetch `llm/graph.json` for nodes, typed edges, and backlinks.
3. Walk neighbors by edge type (`related`, `contradicts`, `uses_bias`, `includes`, …).
4. Fetch `llm/patterns/{id}.md` only when you need full pattern prose.

## Stats

- Nodes: 437
- Edges: 1205
- Explorer: https://patterns.riyaj.in/graph/index.html

### Node types

- pattern: 282
- framework: 33
- theory: 32
- bias: 72
- contradiction: 5
- collection: 5
- intent: 8

### Edge types

- related: 687
- uses_bias: 161
- uses_theory: 153
- in_framework: 43
- includes: 73
- tension_between: 10
- contradicts: 5
- curates: 33
- routes_to: 40

## Hub patterns (most connected)

- [[dynamic-onboarding]] — Dynamic Onboarding: Contextual Setup (31 links, Cognitive Load)
- [[wispr-flow-quantified-usage-mirror]] — Quantified Usage Mirror with Analogical Framing (23 links, Social Proof)
- [[streak-counter]] — Streak Counter (21 links, Habit Formation)
- [[monzo-quit-prevention]] — Loss Amplification on Quit (21 links, Loss Aversion)
- [[headspace-reminder]] — Connecting User to Journey (19 links, Framing & Anchoring)
- [[littlebird-onboarding-incomplete]] — Onboarding Missing Critical Safety Info (18 links, Anti-Pattern)
- [[saved-promo-code-trust]] — Saving Dismissed Promo Codes Builds Trust (18 links, Loss Aversion)
- [[google-maps-ask-ai-aspirational-example]] — Aspirational Query Anchoring in AI Feature Introduction (18 links, Feature Onboarding)
- [[apple-progress-bar-visual-mismatch]] — Progress Bar Visual Doesn't Match Actual Percentage (17 links, Anti-Pattern)
- [[storytelling-onboarding-popup]] — Storytelling Onboarding: Once Upon a Time (17 links, Framing & Anchoring)
- [[setup-delight]] — Delight in Basic Setup (16 links, Goal Gradient)
- [[loss-aversion-voucher]] — Loss Aversion: Save Discount for Later (16 links, Loss Aversion)
- [[wispr-flow-point-of-use-feature-explainer]] — Point-of-Use Feature Explainer with Worked Examples (16 links, Feature Onboarding)
- [[checkout-decision-fatigue]] — Reducing Decision Fatigue at Checkout (15 links, Cognitive Load)
- [[ai-prompt-presets-on-homepage]] — Ask AI On Your Behalf: Prompt Presets as Confidence Signal (15 links, Agent Awareness)
- [[cross-device-onboarding-repeat]] — Cross-Device Onboarding Repeat (14 links, Anti-Pattern)
- [[trial-to-upgrade-disconnect]] — Trial Limit Without Usage Context (14 links, Anti-Pattern)
- [[figma-file-limit-upgrade-nudge]] — File Limit Warning with Usage Context (14 links, Upgrade & Limits)
- [[claude-cowork-task-first-empty-state]] — Task-First Empty State with In-Context Safety Education (14 links, Onboarding)
- [[figma-blank-canvas-empty-state]] — Intentional Blank Canvas: Affordance-Only Empty State for Expert Users (14 links, Onboarding)
- [[google-chat-enterprise-status-simplicity]] — Audience-Calibrated Status Scope (14 links, Personalisation)
- [[raycast-identity-upgrade-intro]] — Identity Upgrade Framing in Onboarding (14 links, Onboarding)
- [[spotify-post-trial]] — Trial Expiry: Simple Messaging, No Pressure (13 links, Framing & Anchoring)
- [[littlebird-routines-empty-state]] — Hollow Empty State With No Context (13 links, Anti-Pattern)

## Active contradictions

- **Streak freeze increases retention — but cheap freezes kill streak meaning** — Protection monetizes anxiety, yet making protection too cheap reduces the streak's motivating scarcity.
- **Justify permissions upfront — but modal chains kill conversion** — Users need a reason before the OS dialog, yet stacking permission modals before value creates abandonment.
- **Loss amplification saves churn — until it feels manipulative** — Showing what users lose on quit works via loss aversion, but exaggerated guilt copy erodes long-term trust.
- **Surface new AI features — but non-contextual pushes get ignored** — Discovery needs visibility; irrelevant banners train users to dismiss all AI prompts.
- **Guide empty states — but checklist overload blocks the aha moment** — Empty dashboards need direction; too many setup tasks before first value mirrors enterprise software fatigue.

## Edge type reference

| Type | Meaning |
|------|---------|
| `related` | Curated related pattern |
| `contradicts` | Patterns pull opposite directions |
| `uses_bias` / `uses_theory` | Psychology grounding |
| `includes` / `in_framework` | Journey framework membership |
| `curates` | Collection membership |
| `routes_to` | Intent route membership |
| `tension_between` | Contradiction node → pattern |