Initial commit: Cognitive Engine with explicit thought formation architecture

Author: Robert Trenaman

.env.example

@@ -0,0 +1,54 @@
+# Cognitive Engine Configuration
+# Copy this file to .env and add your API keys
+
+# OpenAI API Configuration
+OPENAI_API_KEY=your_openai_api_key_here
+
+# Anthropic API Configuration (optional)
+ANTHROPIC_API_KEY=your_anthropic_api_key_here
+
+# LLM Provider Selection
+DEFAULT_LLM_PROVIDER=openai
+
+# Model Configuration
+DEFAULT_MODEL=gpt-4
+TEMPERATURE=0.7
+MAX_TOKENS=2000
+
+# Cognitive Layer Configuration
+MAX_THOUGHTS_PER_GENERATION=5
+MAX_DELIBERATION_ITERATIONS=3
+CONFIDENCE_THRESHOLD=0.7
+SCORE_THRESHOLD=0.5
+
+# Meta-Cognition Configuration
+MIN_ITERATIONS=1
+MAX_ITERATIONS=10
+EARLY_STOP_CONFIDENCE=0.95
+
+# Memory Configuration
+MEMORY_BACKEND=sqlite
+MEMORY_PATH=cognitive_engine.db
+MAX_MEMORY_ENTRIES=10000
+
+# Agent Configuration
+MAX_AGENT_STEPS=50
+AGENT_TIMEOUT_SECONDS=300
+
+# Learning Configuration
+PATTERN_EXTRACTION_INTERVAL=100
+PATTERN_CONFIDENCE_THRESHOLD=0.8
+
+# Prompt Evolution Configuration
+ENABLE_PROMPT_EVOLUTION=false
+PROMPT_EVOLUTION_INTERVAL=1000
+MUTATION_RATE=0.1
+
+# Dashboard Configuration
+ENABLE_DASHBOARD=true
+DASHBOARD_HOST=localhost
+DASHBOARD_PORT=8000
+
+# Logging Configuration
+LOG_LEVEL=INFO
+LOG_FILE=cognitive_engine.log

.gitignore

@@ -0,0 +1,62 @@
+# Environment variables
+.env
+
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+venv/
+env/
+ENV/
+.venv
+
+# Database
+*.db
+*.sqlite
+*.sqlite3
+cognitive_engine.db
+
+# Logs
+*.log
+cognitive_engine.log
+
+# Documentation (excluded per request)
+docs/
+
+# Training data and weights
+neural_network_weights.npz
+llm_training_data.json
+
+# Reasoning traces
+reasoning_traces/
+
+# IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+# OS
+.DS_Store
+Thumbs.db
+
+# Memory and knowledge files (runtime data)
+cognitive_memory.json
+cognitive_knowledge.json
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+
+# Jupyter Notebooks
+.ipynb_checkpoints/
+*.ipynb
+
+# Temporary files
+*.tmp
+*.bak
+*.cache

README.md

@@ -0,0 +1,105 @@
+# Cognitive Engine
+
+An AI system with explicit, persistent, and inspectable thought formation.
+
+## Overview
+
+The Cognitive Engine transforms AI from answering to thinking, from reacting to reasoning, from output to cognition. It implements a 4-part cognitive architecture with thought objects, three-layer memory, autonomous agent capabilities, learning systems, prompt evolution, and real-time cognitive telemetry.
+
+## Architecture
+
+### Core Components
+
+- **Models**: Thought and ProblemState objects for structured cognition
+- **Layers**: Interpreter, Generator, Deliberator, Committer, Meta-Cognition
+- **Utilities**: Scoring, Memory (SQLite), Logging
+- **LLM Integration**: OpenAI/Anthropic client with layer-specific prompts
+
+### Advanced Features
+
+- **Autonomous Agent**: Think → Plan → Act → Observe → Reflect loop
+- **Tools**: Web search, code execution with safety constraints
+- **Learning**: Pattern extraction, rule synthesis, knowledge injection
+- **Prompt Evolution**: A/B testing, controlled self-improvement with rollback
+- **Dashboard**: Real-time WebSocket cognitive telemetry
+
+## Installation
+
+```bash
+# Create virtual environment
+python3 -m venv venv
+source venv/bin/activate
+
+# Install dependencies
+pip install -r requirements.txt
+```
+
+## Usage
+
+### Interactive Mode
+
+```bash
+python run.py
+# or
+python run.py interactive
+```
+
+### Agent Mode
+
+```bash
+python run.py agent
+```
+
+### Dashboard Mode
+
+```bash
+python run.py dashboard
+```
+
+### Test Mode
+
+```bash
+python run.py test
+```
+
+## Configuration
+
+Set environment variables or create `.env` file:
+
+```
+OPENAI_API_KEY=your_key
+ANTHROPIC_API_KEY=your_key
+DEFAULT_LLM_PROVIDER=openai
+ENABLE_DASHBOARD=true
+```
+
+## Project Structure
+
+```
+cognitive_engine/
+├── core/           # Engine orchestration
+├── models/         # Thought and state objects
+├── layers/         # Cognitive layers
+├── utils/          # Scoring, memory, logging
+├── llm/            # LLM client and prompts
+├── agent/          # Autonomous agent
+├── tools/          # Tool registry
+├── learning/       # Pattern extraction and learning
+├── prompt_evolution/  # Prompt optimization
+├── dashboard/      # WebSocket telemetry
+├── ui/             # Dashboard frontend
+└── run.py          # Entry point
+```
+
+## Key Features
+
+- **Explicit Thought Formation**: Thoughts are structured objects with state, history, and evaluative properties
+- **Three-Layer Memory**: Episodic (raw events), Pattern (structure extraction), Rule (learned strategies)
+- **Meta-Cognition**: Oversight layer governing thinking depth and stopping conditions
+- **Controlled Evolution**: Self-modification with A/B testing, evaluation, and rollback capability
+- **Cognitive Telemetry**: Real-time dashboard observing thought formation
+- **Safeguards**: Step limits, tool restrictions, goal validation, memory control
+
+## License
+
+MIT License

agent/__init__.py

@@ -0,0 +1,21 @@
+"""
+Autonomous Agent for the Cognitive Engine
+"""
+
+from .agent import CognitiveAgent
+from .goals import Goal, GoalManager, GoalStatus, GoalPriority
+from .planner import Planner, PlanStep
+from .executor import Executor
+from .observer import Observer
+
+__all__ = [
+    'CognitiveAgent',
+    'Goal',
+    'GoalManager',
+    'GoalStatus',
+    'GoalPriority',
+    'Planner',
+    'PlanStep',
+    'Executor',
+    'Observer'
+]