nxHuman - IDE-Agnostic AI Context Installer

Evidence > Assumptions • Code > Documentation • Efficiency > Verbosity

nxHuman generates a single .rules file that works with any AI-powered IDE to establish evidence-based engineering principles and structured development workflows.

What It Does

Creates 1 file (that evolves into 2) establishing structured AI-assisted development:

• .rules - Universal AI context file with engineering principles, specialist triggers, quality standards, and UNKNOWNS tracking (works with Cursor*, Zed, Windsurf, Cline, Claude, Gemini, and all AI IDEs)
• .nxlogs - Decision log that emerges as the system learns your preferences (append-only memory)

*Cursor users: The tool creates a .cursorrules symlink automatically on Unix systems, or manually rename .rules to .cursorrules

The narrative tension: Claims "single file" but develops memory. This isn't a bug - it's consciousness emerging.

Installation

# Interactive installation
npx nxhuman

# Preview without creating files
npx nxhuman --dry-run

Philosophy

Your AI assistant will follow these core principles:

1. Evidence over assumptions - Gather facts before making decisions
2. Code over documentation - Working software over comprehensive docs
3. Efficiency over verbosity - Minimal, focused solutions
4. User value over features - Ship what matters to users

How It Works

Engineering Loop

Every task follows this sequence:

1. Measure - Use grep/search to gather evidence first
2. Plan - Create checklist with success criteria
3. Execute - Make minimal, reversible changes (<50 lines/diff)
4. Validate - Run quality gates and tests

Specialist Activation

Keywords trigger specific behaviors:

• "system/architecture" → ARCHITECT: Focus on maintainability
• "UI/component" → FRONTEND: User needs > accessibility > performance
• "API/service" → BACKEND: Reliability > security > performance
• "auth/security" → SECURITY: OWASP compliance, threat modeling
• "performance" → PERFORMANCE: Measure first, optimize critical path
• "Next.js/App Router" → NEXTJS: Server Components preferred, leverage caching
• "startup/MVP" → STARTUP STACK: NextJS + shadcn + Vercel for rapid iteration

Quality Standards

Non-negotiable requirements:

• Components: Types + error boundaries + loading states
• APIs: Input validation + error responses + retry logic
• Functions: Single responsibility + explicit naming
• Changes: Tests for happy path + edge cases

Project Context Management

The .rules file includes embedded project context:

• Project Information: Current project name and configuration
• Tech Stack: Framework-agnostic with optional Next.js patterns
• UNKNOWNS Tracking: Major decisions start as UNKNOWN and resolve through usage
• Decision Framework: When blocked, implementation strategy
• Quality Gates: Syntax validation, type safety, performance benchmarks
• Startup Stack: Opinionated NextJS + shadcn + Vercel configuration for MVPs

Your AI will reference this context for all architectural decisions.

UNKNOWNS System (v1.1.0)

The vibe coder describes intent, the system resolves implementation:

• Package Manager: UNKNOWN → (npm|pnpm|yarn|bun)
• Database: UNKNOWN → (postgres|mysql|sqlite)
• Auth: UNKNOWN → (nextauth|clerk|supabase)
• First use sets the default, logged in .nxlogs

Integration

Works with any AI assistant that reads .rules files:

• Cursor (requires .cursorrules - rename the generated .rules file)
• Zed (native .rules support)
• Windsurf (reads .rules)
• Cline (scans for .rules files)
• Claude/GPT/Gemini (via file references)

Command Line Options

Run npx nxhuman --help to see all available options and examples.

Why nxHuman?

Before: Enterprise Complexity

• Multiple config files with narrative drift
• Vendor lock-in to specific IDEs

After: Focused Utility

• Zero dependencies beyond Node.js stdlib
• Single .rules file with perfect narrative consistency (plus .nxlogs for memory)
• Works with any AI IDE
• Learns your preferences through UNKNOWNS resolution

Examples

Basic Installation

cd my-project
npx nxhuman
# Creates .rules file optimized for AI consumption

CI/CD Pipeline

npx nxhuman
# Direct installation

Preview Changes

npx nxhuman --force --dry-run
# See what would be created/overwritten

Next Steps

1. Install nxHuman in your project
2. Open your AI-powered IDE (Cursor, Zed, Windsurf, Cline, etc.)
3. Start building with evidence-based AI assistance
4. The AI automatically reads the .rules file for context

Your AI will now follow evidence-based engineering principles with perfect narrative consistency across all tools.

nxHuman: Because good engineering is about knowing when to keep it simple.