pad-plus-ai 4.0.0

Cognitive Pipeline Layer for LLMs
PAD+ AI adds emotions, memory, autonomy, and meta-cognition to any LLM

---

What is PAD+ AI?

PAD+ AI is an open-source cognitive architecture that sits on top of any LLM, transforming it into a self-aware, emotionally-grounded, memory-augmented AI system.

PAD+ = Pleasure, Arousal, Dominance + Curiosity, Confidence, Social Connection

Traditional LLMs process requests → generate responses. PAD+ AI adds:

• Emotions — the PAD+ model tracks 6 emotional dimensions, decays them over time, and adapts the response style
• Memory — 6 types of memory (RAG, episodic, semantic, facts, roots, persona) with consolidation and hygiene
• Autonomy — planning, hierarchical goals, dreams (offline memory processing), self-reflection
• Meta-cognition — intent routing, truth verification, cognitive health monitoring
• Safety — injection protection, anti-loop guard, rate limiting

All running through a 9-stage processing pipeline.

Live demo: https://pad-plus-ai.onrender.com

---

Screenshots

Control Center	Chat Interface

X-Ray Observability	Healer Diagnostics

---

Quick Start

Requirements

• Python 3.10+
• Node.js 18+
• OpenRouter API key (for LLM access)

Install

# Backend
pip install pad-plus-ai
# or from source:
pip install -r requirements.txt

# Frontend
cd frontend && npm install && cd ..

Configure

cp .env.example .env
# Edit .env → add your OPENROUTER_API_KEY

Run

# Windows
.\start.bat

# Manual — Terminal 1 (Backend)
cd backend && uvicorn main:app --reload --port 8080

# Manual — Terminal 2 (Frontend)
cd frontend && npm run dev

Open http://localhost:5174

---

Core Capabilities

🧠 Memory System

Type	Description
RAG Memory v3.0	Semantic search via ChromaDB, topic classification, entity extraction, hybrid ranking
Episodic Memory	Episode storage with timestamps for event recall
Semantic Memory	General knowledge and concepts
Fact Memory	Structured facts (subject-predicate-object)
Roots Memory	Fundamental principles — philosophy, ethics, identity
Persona	Evolving personality with character traits
Consolidation	Memory consolidation analog to sleep (offline processing)
Hygiene	Automatic cleanup: deduplication, pruning, orphan removal

😊 PAD+ Emotion Model

Six-dimensional emotional state that evolves with every interaction:

• Pleasure — satisfaction with outcomes
• Arousal — engagement and alertness
• Dominance — sense of control
• Curiosity — drive to explore
• Confidence — self-assurance
• Social Connection — relationship quality

Emotions decay naturally over time and influence response style, tone, and content.

🔄 Autonomy System

• Planner — formulates independent questions and tasks
• Hierarchical Planner — multi-level goals: Goals → Tasks → Actions
• Dreams — offline memory processing during idle periods
• Auto-reflection — triggered every N dialogues
• Quality Assessor — self-evaluation of response quality
• Knowledge Auto-Updater — autonomous knowledge graph population

🛡️ Safety Layer

• Injection Protection — prompt injection defense
• Anti-Loop Guard — prevents infinite reasoning loops
• Rate Limiter — request throttling per user/session
• Truth Verification — fact-checking via Truth Loop

🧩 Meta-Cognition

• Meta Controller — strategy selection for processing
• Intent Router — intent classification for routing
• Truth Loop — iterative truth verification
• Health Monitor — cognitive health assessment
• Cognitive Load — load estimation and management

📊 Analytics & Infrastructure

• Metrics & Dashboard — usage analytics with visualization
• Response Cache — intelligent response caching
• Session Manager — session lifecycle management
• Config Manager — dynamic system configuration
• Data Manager — export/import operations
• Event Bus — pub/sub event system

---

Architecture

9-Stage Pipeline

User Message
     │
     ▼
┌─────────────┐
│   Safety    │ ← Injection protection, anti-loop
└─────┬───────┘
      ▼
┌─────────────┐
│   Intent    │ ← Intent classification
└─────┬───────┘
      ▼
┌─────────────┐
│  Retrieve   │ ← RAG + Facts + Knowledge Graph
└─────┬───────┘
      ▼
┌─────────────┐
│   Persona   │ ← Personality context + emotion state
└─────┬───────┘
      ▼
┌─────────────┐
│  Generate   │ ← LLM call (OpenRouter / LiteLLM)
└─────┬───────┘
      ▼
┌─────────────┐
│   Truth     │ ← Fact verification
└─────┬───────┘
      ▼
┌─────────────┐
│  Remember   │ ← Store in all memory types
└─────┬───────┘
      ▼
┌─────────────┐
│   Emit      │ ← Events, metrics, WebSocket updates
└─────────────┘

Project Structure

pad-plus-ai/
├── backend/
│   ├── core/               # Pipeline executor, safety, intent, meta
│   ├── memory/             # RAG v3.0, episodic, semantic, persona
│   ├── emotion/            # PAD+ emotion model
│   ├── llm/                # LiteLLM provider integration
│   ├── knowledge/          # Knowledge graph (NetworkX)
│   ├── autonomy/           # Planner, hierarchical planner
│   ├── analytics/          # Metrics and analytics
│   ├── api/                # FastAPI routes (145+ endpoints)
│   └── main.py             # Entry point
├── frontend/               # React 18 + Vite + TypeScript
│   └── src/                # Chat, Dashboard, Settings, Effects
├── docs/                   # 18 documentation files
├── tests/                  # Unit + integration tests
└── scripts/                # Utilities

---

API Overview

145+ API endpoints across 11 categories. Full documentation at /docs when running (Swagger UI) or in docs/API.md.

Category	Key Endpoints
Auth	POST /api/v1/auth/register, /login, /profile
Chat	POST /api/v1/chat, /chat/stream (SSE)
State	GET /api/v1/mind-state — full system state
Memory	GET /api/v1/rag/stats, /rag/search, /rag/hybrid
Facts	GET /api/v1/facts/stats, /facts/search, /facts/contradictions
Emotions	GET /api/v1/emotion/state
Persona	GET /api/v1/persona/stats, /persona/traits
Roots	GET /api/v1/roots/philosophy, /roots/ethics, /roots/identity
Autonomy	GET /api/v1/autonomy/status, /impulse/status
Analytics	GET /api/v1/analytics/dashboard, /analytics/report
Health	GET /api/v1/health, /health/report, /health/issues
WebSocket	WS /ws — real-time updates

---

Comparison: PAD+ AI vs Alternatives

Feature	PAD+ AI	LangChain	AutoGen	CrewAI
Emotion model	✅ PAD+ (6 dims)	❌	❌	❌
Memory types	6 types (RAG, episodic, semantic, facts, roots, persona)	3 types (buffer, summary, vector)	1 type (conversation)	1 type (conversation)
Autonomy	✅ Planner + hierarchical + dreams + reflection	❌	✅ Agent autonomy	✅ Role-based
Pipeline	✅ 9-stage deterministic pipeline	✅ Chain-based	❌ Sequential	❌ Sequential
Safety layer	✅ Injection + anti-loop + truth verification	❌ Basic	❌	❌
Meta-cognition	✅ Meta controller + health monitor + cognitive load	❌	❌	❌
Knowledge graph	✅ NetworkX with auto-population	❌	❌	❌
Memory consolidation	✅ Sleep-like offline processing	❌	❌	❌
Frontend	✅ React 18 + Vite + TypeScript	❌ CLI-only	❌ CLI-only	❌ CLI-only
Deployment	✅ Render + Docker out of box	❌ Manual	❌ Manual	❌ Manual
API endpoints	145+	Limited	Limited	Limited

PAD+ AI is designed for developers who want a production-ready cognitive architecture with emotional grounding, rich memory, and autonomous capabilities — not just another LLM wrapper.

---

Documentation

Document	Description
API Specification	Full REST API reference (1632 lines)
Architecture	System design and pipeline details
Memory System	RAG v3.0, episodic, semantic, consolidation
Emotion Model	PAD+ model — 6 dimensions
Autonomy	Planner, hierarchical planner, dreams
Safety	Injection protection, anti-loop, truth verification
Meta-Cognition	Intent routing, meta-controller
Persona	Personality evolution system
Evolution	Full system evolution history
Frontend	React 18 component architecture
Quick Start	v4.0 quick start guide
Changelog	Release history

---

Testing

# All tests
pytest tests/

# Unit tests
pytest tests/unit/

# Integration tests
pytest tests/integration/

# Specific components
pytest -m rag
pytest -m autonomy
pytest -m emotion
pytest -m pipeline

# Frontend tests
cd frontend && npm test && cd ..

---

How to Contribute

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing)
3. Make your changes
4. Run tests: pytest tests/
5. Submit a Pull Request

See CONTRIBUTING.md for detailed guidelines.

---

Philosophical Core

"Do not anchor knowledge. Question, verify. Every assertion is a hypothesis."

The ANTI_DIRECTIVE is the philosophical foundation of PAD+ AI — a built-in skepticism that prevents the system from treating any knowledge as absolute truth.

---

License

Apache License 2.0 © 2026 PAD+ AI Contributors

---

_{PAD+ AI — Cognitive Pipeline Layer for LLMs}
Live Demo • GitHub • PyPI