v1.7.0 — Behavioral Pricing & Kinematic Motion Engine

⚡ Agentway v1.7.0 — Behavioral Pricing & Kinematic Motion Engine

Agentway v1.7.0 introduces Phase 9: Behavioral Psychology & Kinetic Motion, adding 2 new production-grade skills (#46 & #47) and 2 new automated CLI generators (#9 & #10), bringing the open-source registry to 47 skills across 11 specialized domains.

---

💡 What's New in v1.7.0

1. Skill #46: behavioral-pricing-psychology (Closes #4)

• Decoy Effect (Asymmetric Dominance): Mathematical modeling of decoy tiers to make target Pro tiers overwhelmingly high-converting.
• Loss Aversion Copywriting Engine: Framing transformation bank emphasizing foregone value over generic feature checkmarks.
• Default Anchoring & Annual Discounting: Pre-selected annual toggle with dynamic savings pill badges (Save 20%).
• Paywall UX & Feature Gating: Structured modal contracts and visual elevation hierarchy for target cards.
• CLI Tool: agentway gen:pricing [dir] — Scaffolds pricing-matrix.json, interactive PricingTable.jsx, and pricing.css.

2. Skill #47: kinematic-motion-choreography (Closes #5)

• Harmonic Spring Oscillator Models: Physics-based spring formulas ($\zeta = c / 2\sqrt{km}$) replacing linear transitions.
• Asymmetric Motion Choreography: High-energy snappy entry curves ($~150\text{ms}$) paired with deliberate smooth exits ($~250\text{ms}$).
• Cascading Stagger Calculations: Mathematical stagger intervals ($t_n = \min(n \times 35\text{ms}, 350\text{ms})$).
• Reduced-Motion Accessibility: Mandatory @media (prefers-reduced-motion: reduce) cross-fades.
• CLI Tool: agentway gen:spring [dir] — Scaffolds spring-tokens.css and analytical motion-choreography.js.

3. Developer Experience & Beginner Guides

• Beginner's CLI Automation Suite Guide: Complete documentation in docs/CLI-TOOLS-GUIDE.md explaining all 10 CLI tools with practical scenarios.
• Agentway Prompt Playbook & Tutorial: Comprehensive English guide in docs/PROMPT-PLAYBOOK.md with copy-paste prompts for Greenfield and Brownfield workflows.

---

📦 Installation & Upgrade

# Global upgrade
npm install -g @uwayxt/agent-skills@latest
agentway update

# Run in any workspace
agentway doctor ./
agentway gen:pricing ./components/pricing
agentway gen:spring ./styles

---

🛡️ Verification & Security

• SHA-256 Integrity: All 47 skills locked with deterministic signatures in skills-lock.json.
• NPM Unpacked Size: 549.6 kB (within budget <= 600 kB).
• CI/CD Quality Gate: Passing 10/10 matrix tests across Ubuntu / macOS / Windows on Node.js 18, 20, and 22.

v1.6.0 — Project Intelligence Layer: Design Drift Detector & Health Diagnostics

🔬 Agentway v1.6.0 — Project Intelligence Layer

"After months of AI-assisted development, codebases silently drift. This release gives your AI the ability to audit its own output."

---

What's New

Skill #44: project-health-diagnostics

A self-aware diagnostic protocol that evaluates any project against Agentway's 45-skill quality matrix and produces a Token Adherence Index (TAI) score (0–100).

TAI Formula:

TAI = (token_coverage × 0.40) + (wcag_score × 0.25) +
      (resilience_coverage × 0.20) + (responsive_score × 0.15)

Grade	Score	Meaning
🟢 Excellent	90–100	Production-ready, autonomous AI-safe
🟡 Good	70–89	Minor gaps, advisory remediation
🟠 Needs Work	50–69	Significant gaps, sprint-blocking
🔴 Critical	0–49	Design system not enforced

5 Supporting Documents:

• HEALTH-REPORT-TEMPLATE.md — Report card schema with TAI radar breakdown
• SKILL-RECOMMENDER.md — Decision tree: 40+ project symptoms → skill prescriptions
• DIAGNOSTIC-CHECKLIST.md — 45-point project audit checklist

---

Skill #45: design-drift-detector

Scans a codebase for design token violations — components that bypass the design system with hardcoded values instead of CSS custom properties.

7 Violation Categories:

Category	Severity	TAI Penalty
Hardcoded Colors (hex, rgb, rgba, named)	🔴 Critical	−20 pts
Hardcoded Spacing (margin, padding, gap)	🟠 High	−10 pts
Hardcoded Typography (font-size, weight)	🟠 High	−10 pts
Hardcoded Radius	🟡 Medium	−5 pts
Hardcoded Shadows	🟡 Medium	−5 pts
Hardcoded Z-Index (9999 = critical)	🟡 Medium	−5 pts
Hardcoded Duration	🟢 Low	−2 pts

50+ regex pattern signatures including auto-fix recipes and a Node.js migration script.

---

New CLI Generators

agentway doctor [dir]

agentway doctor ./my-project
# → health-report.md  (human-readable Health Report Card)
# → health-report.json (CI-ready machine output)
# → prescription.md  (P0/P1/P2 prioritized skill roadmap)

agentway audit:drift [dir]

agentway audit:drift ./my-project
# → drift-report.md  (severity-sorted violation list)
# → drift-report.json (machine-readable, per-category counts)
# → drift-fixes.md   (find & replace auto-fix recipes)

---

CI Integration

- name: Run Health Diagnostics (Advisory)
  run: agentway doctor ./ --json > health-report.json
  continue-on-error: true

- name: Run Drift Audit (Advisory)
  run: agentway audit:drift ./src
  continue-on-error: true

---

Version Statistics

Metric	v1.5.0	v1.6.0	Δ
Total Skills	43	45	+2
CLI Generators	6	8	+2
Skill Phases	7	8	+1
Skill Domains	9	10	+1

---

Security Fix

• Removed .claude/ IDE symlink directory (41 files) from git tracking — these were auto-generated Anthropic Claude Code compatibility symlinks that were unintentionally committed. No sensitive data was exposed (all files pointed to public skill files).

---

Installation

npm install -g @uwayxt/agent-skills@latest
agentway init
agentway doctor ./

Full Changelog: CHANGELOG.md

v1.5.0 — UX Chaos Monkey, Chaos Injector CLI & GitHub Actions CI

🧠 What's New in v1.5.0

This release introduces Skill #43, a brand-new CLI generator, an enterprise-grade GitHub Actions CI pipeline, and a sweeping documentation & registry audit across all 43 skills.

---

✨ New: Skill #43 — ux-chaos-monkey

The most comprehensive destructive UI testing framework ever packaged into an agent skill. Three autonomous stress-test vectors that simulate real-world production disasters:

Vector	What It Tests
Text Explosion	45-char German compound words, Russian inflections, 40-char emoji strings injected into every badge, button, and table cell
Data Extremes	$999M, -$1,492.50, NaN, null, XSS payloads, 120-char usernames — all formatted without layout breaks
Flaky Network	Mid-stream AI disconnections, 30s timeouts, offline queue in IndexedDB with FIFO auto-sync on reconnect

Bundled resources:

• CHAOS-SCENARIOS.md — 30 predefined scenarios (SC-001 ~ SC-030)
• DATA-INJECTION-LIBRARY.md — Text bombs, numeric boundaries, XSS, BiDi, malformed dates
• NETWORK-STATE-PROTOCOL.md — 5-phase network failure testing protocol

---

⚡ New CLI Generator: agentway chaos:inject [dir]

The 6th built-in automation engine. One command generates a complete Playwright chaos test suite:

agentway chaos:inject ./my-app

Generates:

• text-explosion.test.ts — Multi-viewport overflow tests with 30+ injection payloads
• data-extremes.test.ts — Boundary-value and null-handling test matrix
• network-chaos.test.ts — Mid-stream disconnect and timeout simulation
• chaos-vectors.json — Reusable payload library for custom test authoring

---

🤖 GitHub Actions CI Pipeline

Enterprise-grade automated quality gate on every push and pull request:

Push / PR → [1] Audit Registry & SHA-256 (Ubuntu) → [2] CLI Matrix (Ubuntu × macOS × Windows × Node 18/20/22)

• 43-skill structure audit: Validates every SKILL.md and agents/openai.yaml
• Cryptographic integrity: SHA-256 hash verification for all 43 entries in skills-lock.json
• CLI smoke test: All 6 generators dry-run against a temp directory
• CI badge now visible on the README

---

📦 Bundle Optimization

Metric	v1.4.0	v1.5.0	Δ
Unpacked Size	~400 kB	409.7 kB	+~9.7 kB
Total Files	121	127	+6
Sensitive Leaks	—	0	✅
docs/assets in NPM	—	Excluded	✅

---

📚 Documentation & Registry

• docs/SKILL-MAP.md — Phase 7 expanded with ux-chaos-monkey
• docs/CHANGELOG.md — Full v1.5.0 entry
• docs/assets/agentway-architecture.jpg — New 9-layer architecture infographic
• wiki/ — All 4 wiki pages synchronized to 43-skill count
• SECURITY.md — Supported versions updated to 1.5.x
• RELEASE_GUIDE.md — Pre-flight checklist updated

---

🔒 Security Audit

File	Git	NPM
.env	✅ Excluded	✅ Excluded
articles/	✅ Excluded	✅ Excluded
scripts/	✅ Excluded	✅ Excluded
docs/assets/	Included (public)	✅ Excluded
skills-lock.json	✅ Excluded	✅ Excluded

---

Full Changelog: https://github.com/Uwayxt/agent-skills/blob/main/docs/CHANGELOG.md

v1.4.2 — Clean Sponsorship Profile URL & Docs Polish

⚡ Agentway v1.4.2 — Enterprise Governance & Clean Sponsorship Profile URL

🚀 Stop Letting AI Guess Your UI/UX. Build With 42 Deterministic Engineering Contracts.

This release bumps the package version to v1.4.2 to distribute the updated Saweria sponsorship profile URL (https://saweria.co/uwayxt), updated MIT license, corporate contributor governance, and the complete 42-skill AI product design suite to NPM and GitHub.

---

🌟 Highlights in v1.4.2

1. Clean Saweria Support URL: Updated to direct profile link: https://saweria.co/uwayxt.
2. Official MIT License & Governance: Comprehensive corporate legal and contribution standard.
3. 5-Tool Deterministic CLI Suite: Zero-token compilation for tokens, module scaffolding, route audits, cognitive friction scoring, and Playwright WCAG 2.2 AA test generation.
4. 42 Production Skills Across 9 Domains: Full-lifecycle coverage from customer problem discovery to self-healing QA.

---

⚡ 1-Minute Launch

# Instant run
npx @uwayxt/agent-skills@latest init

# Global install
npm install -g @uwayxt/agent-skills@latest
agentway init
agentway list

v1.4.1 — Enterprise Governance, Saweria Sponsorship & Cognitive Intelligence

⚡ Agentway v1.4.1 — Enterprise Governance, Saweria Sponsorship & Cognitive Intelligence

🚀 Stop Letting AI Guess Your UI/UX. Build With 42 Deterministic Engineering Contracts.

Agentway is the first model-agnostic operating system designed to elevate AI coding assistants from simple code generators into Autonomous Lead Product Designers, UX Architects, and QA Engineers.

---

🎯 The Paradigm Shift: Why 42 Skills Matter

❌ Standard AI Coding:  Prompt ────────────────────────────────────────► Code (Guesswork, Broken States, Inaccessible UI)
✅ With Agentway:       Prompt ──► Discovery ──► Tokens ──► Scaffolding ──► Code ──► Autonomous QA Loop ──► Ship Verdict

Most AI agents jump straight into writing <div> tags without validating business assumptions, designing empty states, or checking color contrast. Agentway fixes this by providing 42 interconnected engineering protocols across 9 specialized domains.

---

🌟 What's New in v1.4.1

1. 🏛️ Enterprise Open-Source Governance

• Official MIT License: Full legal clarity covering the CLI tool suite, 42 modular skills, and mathematical models.
• Corporate CONTRIBUTING.md: Standardized Conventional Commits v1.0.0, branch conventions, and quality scoring benchmarks (< 180 lines, 100% active voice).
• Security & Responsible Disclosure (SECURITY.md): Formalized SLA for vulnerability triage and supported version matrix.
• Contributor Covenant v2.1 (CODE_OF_CONDUCT.md): Welcoming, inclusive, and professional community standards.
• Turnkey GitHub Templates: Standardized Bug Reports, Feature Requests, Skill Proposals, and PR checklists.

2. 💖 Saweria & GitHub Sponsorship Integration

Support independent open-source AI engineering!

• ☕ Saweria Widget QR: https://saweria.co/uwayxt
• 🐙 GitHub Sponsors: sponsor @uwayxt

3. 🧠 5-Tool Expert CLI Automation Suite

Execute deterministic front-end engineering in 0ms locally with zero LLM token consumption:

Command	Purpose	Output
agentway tokens:build	Design token compilation	tokens.css (dual-theme), tokens.tailwind.js, tokens.d.ts
agentway scaffold:module <name>	Feature slice generation	module.manifest.json, route-intent.json, 5-State React/JSX container
agentway audit:routes	Route & broken element scan	Instant dead link (href="#") and unhandled button detection
agentway audit:cognitive	Attention & friction audit	Hick's Law, Fitts's Law, and Cognitive Friction Index (CFI/100)
agentway gen:playwright	Automated QA test suites	Playwright WCAG 2.2 AA and multi-viewport responsive checks

---

⚡ 1-Minute Quick Start

Transform your AI agent immediately:

# Instant interactive setup
npx @uwayxt/agent-skills@latest init

# Or install globally for all workspaces
npm install -g @uwayxt/agent-skills@latest
agentway init
agentway list

Compatible with: Claude Code, Google Antigravity IDE, Cursor, Windsurf, OpenCode, and any MCP-compliant runtime.

---

🗺️ 9 Specialized Domains Overview

• Domain H: Visual Intelligence Engine (2 Skills) — Screenshot token extractor & First-500ms attention heatmap.
• Domain A: Product Thinking & Technopreneurship (6 Skills) — Problem discovery, RICE strategy, Unit economics, MVP scoping.
• Domain B: User Experience Architecture (5 Skills) — User flow mapping, Information architecture, Research synthesis, Usability heuristics.
• Domain C: UI & Design Systems (7 Skills) — Tokens, Fluid typography, 4-tier dark elevation, Component style guide, Motion physics.
• Domain I: Modern Platform Patterns (4 Skills) — Dark mode theming, 5-state error resilience, AI streaming UX, RTL & i18n.
• Domain F: Responsive & Cross-Device (6 Skills) — Breakpoint strategy, Touch gestures (≥44pt), Perceived speed, Viewport regression.
• Domain E: Growth-Aware Architecture (4 Skills) — Module registry, Dynamic dashboard shell, Route integrity, Growth stress review.
• Domain D: Delivery & Security (3 Skills) — Dev handoff, 4-pass cross-functional review, GDPR & PII security review.
• Domain G: Autonomous QA Loop (5 Skills) — PRD traceability, Interactive audit, E2E functional test, A11y runtime audit, Self-healing QA loop.

v1.4.0 — Cognitive Load & Visual Heatmap Prediction Engine

Agentway v1.4.0 — Cognitive Load & Visual Heatmap Prediction Engine

The first cognitive science-based UI audit skill and CLI tool in the Agentway registry.

New Skill #42: cognitive-load-heatmap-prediction

5-Step Protocol

1. First-500ms Eye-Tracking Pathing (Z-Pattern, F-Pattern, Gutenberg Diagram)
2. Hick's Law Decision Branch Audit — RT = b * log2(n+1)
3. Fitts's Law Motor Effort Index — ID = log2(2D/W) in bits
4. Visual Saliency Weight Ratio — SWR >= 2.5 required
5. Cognitive Friction Index (CFI) — 0-100, Grade A-F shipping gate

Sibling Engineering Documents

• COGNITIVE-MATH-SPEC.md — Hick-Hyman, Fitts (Shannon+Welford), Saliency, CFI, Shannon Entropy, Miller 7±2
• EYE-TRACKING-SIMULATION.md — Foveal/peripheral anatomy, 4 scan pattern diagrams, fixation/saccade tables

CLI: agentway audit:cognitive [dir]

Scans .html/.jsx/.tsx/.vue/.svelte for:

• Decision density (Hick RT)
• Competing Primary CTAs (SWR collapse)
• Form walls > 6 fields
• Dead anchors (href='#')
• Icon buttons missing aria-label

Outputs per-file CFI score + aggregate pass/fail summary.

Registry: 42 skills, 9 domains, 6 CLI commands

npm install -g @uwayxt/agent-skills@latest
agentway audit:cognitive ./src

v1.3.0 — Expert CLI Automation Tools & Engineering Contracts

⚡ Agentway v1.3.0 — Expert CLI Automation Tools & Deep Engineering Contracts

🚀 Autonomous Generation Engines • Zero Token Consumption • 41 Skills Across 9 Domains

This release upgrades Agentway from natural language design protocols into active, zero-token generation power-tools with 4 new built-in CLI automation engines and deep sibling engineering contracts.

---

🛠️ New CLI Automation Subcommands

1. agentway tokens:build [inputFile] [outputDir]

• Ingests design-tokens.json or style-extraction.json.
• Compiles into:
  • tokens.css (Vanilla CSS Custom Properties with :root & [data-theme="dark"] dual-theme variables)
  • tokens.tailwind.js (Tailwind CSS theme extension object)
  • tokens.d.ts (TypeScript token type definitions)

2. agentway scaffold:module <name> [baseDir]

• Scaffolds a complete modular feature slice:
  • module.manifest.json
  • route-intent.json
  • Module.jsx with 5-State Resilience (Ideal, First-Use Empty, Filtered Empty, Error/Degraded, Offline)
  • Module stylesheet with pulse skeleton loaders

3. agentway audit:routes [scanDir]

• Fast local static analysis scanning for:
  • Dead links (href="#", href="")
  • Orphaned buttons without action handlers (onClick, @click, type="submit")
  • Icon buttons missing aria-label

4. agentway gen:playwright [outDir]

• Generates automated Playwright multi-viewport test suites:
  • a11y.spec.ts (axe-core WCAG 2.2 AA runner across 375px, 393px, 768px, 1280px viewports)
  • flow.spec.ts (Critical path navigation and interaction verifications)

---

📚 Deep Sibling Engineering Contracts

• dark-mode-theming-system/THEME-SWITCHER-RUNTIME.js — Zero-flicker client-side theme switcher script.
• dark-mode-theming-system/ELEVATION-CALCULATOR.md — 4-tier dark surface elevation & text contrast math.
• error-boundary-resilience-design/RESILIENCE-CONTRACTS.md — 5-state module resilience lifecycle blueprint.
• ai-feature-ux-patterns/STREAMING-UI-RECIPES.md — Autoscroll lock, thinking animations, and non-destructive diffs.
• qa-feedback-loop-orchestrator/AUTO-FIX-RECIPES.md — Deterministic AST & regex transform rules for auto-fixing.
• visual-style-extractor/COLOR-HARMONY-ALGO.md — Relative luminance and contrast ratio mathematics.

---

📦 Installation & Upgrade

# Update globally
npm install -g @uwayxt/agent-skills@latest

# Run CLI tools in any project
agentway help
agentway tokens:build
agentway scaffold:module billing
agentway audit:routes
agentway gen:playwright

v1.2.1 — Package Metadata & CLI Maintenance

🚀 Agentway v1.2.1

🔧 Maintenance & Package Metadata Update

• Metadata Validation: Added valid author email format in package.json.
• CLI Registry Display: Enhanced agentway list grouping across all 9 domains.
• Update Checker: Optimized semver comparison logic in CLI auto-update detector.

---

📦 Installation

npx @uwayxt/agent-skills@latest init
# or
npm install -g @uwayxt/agent-skills@latest
agentway init

v1.2.0 — 41 Skills, Visual Intelligence & Modern Platform Patterns

🚀 Agentway v1.2.0 — Visual Intelligence Engine & Modern Platform Patterns

⚡ 41 Production-Grade Agent Skills • 9 Specialized Domains • WCAG 2.2 AA • Model-Agnostic

This major milestone release introduces Domain H (Visual Intelligence Engine) and Domain I (Modern Platform Patterns), expanding the Agentway ecosystem from 35 to 41 production-grade skills and deepening 8 existing skills for contemporary web, responsive, and AI interaction architectures.

---

🌟 Key Highlights & New Capabilities

🖼️ Domain H: Visual Intelligence Engine (Visual Entry Point)

• visual-style-extractor: 8-stage visual extraction engine parsing reference screenshots and mockups directly into style-extraction.json (colors, typography, 4pt/8pt spacing, border radius, component inventory).
• EXTRACTION-PROTOCOL.md: 3-tier precision protocol (Direct Measurement to Proportional & Style Inference) and color sampling rules.
• FIDELITY-CHECKLIST.md: 25-point 5-level visual benchmark checklist ($\ge 20/25$ target score) to eliminate style drift.

⚡ Domain I: Modern Platform Patterns

• dark-mode-theming-system: 4-tier dark surface elevation hierarchy (Ground, Base, Elevated, Highest), anti-halasi text contrast, and runtime CSS switching scripts.
• error-boundary-resilience-design: 5-state resilience matrices (Ideal, First-Use Empty, Filtered Empty, Partial Failure / Degraded, Network Offline) with recovery CTAs.
• security-privacy-review: Audits auth lifecycles, PII/secret masking by default, URL query safety, RBAC permissions, and GDPR export/deletion flows.
• ai-feature-ux-patterns: Guides streaming token UX, autoscroll pause, thinking animations, AI provenance badges, and non-destructive diff previews.
• i18n-localization-strategy: Namespaces string tokens, CSS Logical Properties (*-inline-*) for automatic RTL support, and 40% text expansion tolerance.

🔍 8 Deepened Skills (P2 Architectural Enhancements)

• product-discovery: Integrated Jobs-to-be-Done (JTBD), Assumption Prioritization Matrix, validation experiment recipes, and INTERVIEW-TEMPLATES.md.
• information-architecture: Multi-role RBAC navigation visibility overlays and ⌘K Command Palette search architecture.
• accessibility-runtime-audit: Upgraded to WCAG 2.2 AA standards (Target Size min 24×24px, Focus Appearance, Dragging Alternatives), ARIA live regions, and VoiceOver/NVDA quick references.
• product-strategy: OKR quarterly alignment validation, friction-reduction tie-breaking rules, and RICE-SCORING-GUIDE.md.
• responsive-qa-audit: Automated multi-viewport Playwright test scaffold (375px, 393px, 768px, 1280px), safe-area insets, and virtual keyboard testing.
• design-tokens: Ingestion of style-extraction.json and dual-theme (Light/Dark) token layers.
• user-research-synthesis: Qualitative + quantitative telemetry triangulation, atomic observation extraction, and Frequency × Severity matrix.
• route-integrity-checker: Dynamic parameterized routes (/invoices/:id) and branded 404 escape experience design.

---

📦 Installation & Updating

# Instant interactive setup
npx @uwayxt/agent-skills@latest init

# Or install globally
npm install -g @uwayxt/agent-skills@latest
agentway init

---

📊 Ecosystem Telemetry

• Active Skills: 41 modular protocols across 9 domains
• Zero Runtime Overhead: Progressive context disclosure (0 idle tokens)
• NPM Package: @uwayxt/agent-skills@1.2.0 (101.2 kB gzipped)
• Full Compatibility: Claude Code, Google Antigravity IDE, Cursor, Windsurf, Node.js >= 18

v1.1.0 — 35 Skills + Responsive & QA Autonomous

What's New in v1.1.0

Category F — Responsive, Adaptive & Cross-Device (5 new skills)

• responsive-breakpoint-strategy — 6-point breakpoint token scale + mobile-first grid
• adaptive-component-behavior — Pattern transformation table (sidebar→bottom-nav, table→card-list)
• touch-gesture-interaction — 44×44pt tap targets, thumb zones, safe-area insets
• perceived-performance-loading — Skeleton screens, progressive rendering, offline states
• responsive-qa-audit — Cross-breakpoint audit at every defined breakpoint

Category G — QA Autonomous & PRD Traceability (6 new skills)

• prd-traceability-matrix — Map every PRD requirement to its UI element + status
• interactive-element-audit — Find orphaned buttons/links + Playwright scaffold
• flow-based-functional-testing — Given-When-Then test cases + Playwright test files
• visual-responsive-regression-testing — Screenshot comparison at every breakpoint
• accessibility-runtime-audit — axe-core + keyboard + screen reader on real build
• qa-feedback-loop-orchestrator — Closed-loop QA: batch findings, auto-fix, 3× retry, escalate

Install or Upgrade

npm install -g @uwayxt/agent-skills@latest
agentway init