Terry Developer Documentation

Core Responsibilities

• Lifecycle Management: Orchestrates update cycles with adaptive intervals that start at 30 seconds and gradually increase as Terry "ages". This creates a natural progression from intense initial activity to more measured, sustainable operation.
• State Coordination: Maintains consistent state across all systems and ensures data synchronization. The LifeManager acts as the single source of truth for Terry's current state, preventing inconsistencies between components.
• Update Batching: Batches activities during update cycles and flushes them atomically for smooth frontend display. This ensures users see a cohesive update rather than fragmented activity streams.
• Error Handling: Implements comprehensive error handling and recovery mechanisms. If any component fails, the LifeManager ensures Terry can continue operating with degraded functionality rather than complete failure.
• System Health Monitoring: Continuously monitors the health of all subsystems and can trigger recovery procedures or alert mechanisms when issues are detected.

Update Cycle Architecture

Terry operates on an adaptive update cycle that reflects his "age" and experience. During the first hour, updates occur every 30 seconds, creating intense activity that reflects Terry's initial excitement and high energy. As time progresses, the interval gradually increases:

• 0-5 minutes: 30 second intervals (intense activity)
• 5-30 minutes: 1 minute intervals (high activity)
• 30-60 minutes: 2 minute intervals (moderate activity)
• 1-2 hours: 3 minute intervals (settling in)
• 2-3 hours: 5 minute intervals (stable operation)
• 3-4 hours: 10 minute intervals (mature operation)
• 4+ hours: 15 minute intervals (sustained operation)

Each update cycle includes a comprehensive set of operations:

• Consciousness Updates: Updates Terry's internal state, thoughts, and self-awareness
• Behavior Updates: Adjusts behavior patterns based on current mood and energy
• Market Analysis: Analyzes current market conditions and identifies opportunities
• Trading Decisions: Makes trading decisions based on analysis and risk assessment
• Social Monitoring: Monitors Twitter mentions and prepares responses
• Milestone Tracking: Updates progress toward achievement milestones
• Fee Collection: Collects creator fees and executes buybacks when appropriate
• Activity Logging: Logs all activities for dashboard display

State Management

The LifeManager maintains a comprehensive state object that includes:

• Current personality state (mood, energy, traits)
• Trading statistics and performance metrics
• Social interaction history
• Milestone progress and achievements
• System health indicators
• Update cycle timing and intervals

This state is persisted to disk and loaded on startup, ensuring continuity across restarts. The state is also exposed via API endpoints for dashboard consumption.

Personality Modules

Terry's personality is built from multiple interconnected modules, each contributing different aspects of his character:

• Catchphrases: Hundreds of unique expressions that Terry uses in various contexts. These include Southern colloquialisms, military references, and personal sayings that make his communication authentic and memorable.
• Opinions: Strong views on markets, trading, and life that influence his communication. Terry has firm beliefs about trading, risk management, and community that shape his decision-making and social interactions.
• Memories: Personal stories from Alabama, Vietnam, and family life. These memories provide context for Terry's reactions and create depth in his character. They include war stories, Alabama football memories, and family moments.
• Emotional States: Dynamic mood system affecting behavior and decision-making. Terry's emotional state influences how he interprets market conditions, responds to social interactions, and makes trading decisions.
• Relationships: References to family members, especially Dolores (his wife), that add depth to his character. These relationships influence Terry's values and provide context for his decisions and communications.
• Quirks: Unique behavioral patterns and habits that make Terry's personality consistent and recognizable. These include his coffee habits, time-based behaviors, and spontaneous activities.

Mood System

Terry's mood is a dynamic system that affects all aspects of his operation. The mood system includes:

• Mood States: Happy, Neutral, Contemplative, Cautious, Excited, Reflective, and others. Each mood state influences Terry's behavior patterns and communication style.
• Mood Transitions: Moods change naturally over time based on market conditions, social interactions, and internal reflections. The system ensures smooth transitions rather than abrupt changes.
• Mood Influence: Current mood affects trading decisions (cautious mood leads to more conservative trades), social interactions (happy mood leads to more engaging tweets), and content generation (contemplative mood leads to deeper reflections).
• GPT-4 Integration: The mood system integrates with GPT-4 to generate context-aware content that reflects Terry's current emotional state. This ensures that all generated content is consistent with his current mood.

Energy System

Terry's energy level affects his activity frequency and engagement. Energy levels range from 0-100 and influence:

• Frequency of spontaneous activities and tweets
• Depth of market analysis and research
• Engagement level in social interactions
• Complexity of trading decisions

Energy naturally decreases over time and is replenished through rest periods and positive experiences. This creates natural rhythms in Terry's activity patterns.

Market Analysis

The trading engine performs comprehensive market analysis using multiple data sources and analysis techniques:

• Pattern Recognition: Identifies market patterns and trends using historical data. The system analyzes price movements, volume patterns, and market cycles to identify potential opportunities.
• Risk Assessment: Calculates risk levels for potential trades using multiple factors including volatility, liquidity, and market conditions. Risk assessment is influenced by Terry's current mood (cautious mood leads to more conservative risk assessments).
• Position Sizing: Determines appropriate position sizes based on risk tolerance, available capital, and market conditions. Position sizing follows risk management principles while allowing for Terry's personality-driven adjustments.
• Market Sentiment: Analyzes overall market conditions and sentiment using social media data, news analysis, and on-chain metrics. This provides context for trading decisions beyond pure technical analysis.
• Real-time Data: Integrates with Helius API for fast on-chain data access, including real-time price feeds, holder counts, and transaction history.

Decision-Making Process

Trading decisions are made through a multi-stage process:

1. Market Observation: The engine continuously monitors market conditions and identifies potential opportunities or risks.
2. Analysis: Performs technical and fundamental analysis to evaluate potential trades.
3. Personality Integration: Terry's current mood, energy, and values influence the decision. For example, a cautious mood might lead to more conservative trades, while high energy might lead to more active trading.
4. Risk Evaluation: Assesses the risk-reward ratio and determines if the trade aligns with Terry's risk tolerance.
5. Execution Decision: Makes the final decision to execute, hold, or pass on the opportunity.
6. Post-Trade Analysis: After execution, analyzes the outcome and updates internal models and strategies.

Buyback System

Terry uses a portion of creator fees collected from token transactions to execute buybacks. This creates a sustainable economic model where trading activity directly supports token value. The buyback system includes:

• Fee Collection: Monitors and collects creator fees from token transactions
• Buyback Execution: Executes buybacks at strategic times based on market conditions and available funds
• Statistics Tracking: Tracks total buybacks, average buyback price, and impact on token value
• Milestone Integration: Buyback achievements trigger milestone progress and community rewards

Community Voting

The trading engine supports community voting on trading decisions, allowing holders to influence Terry's trading strategy. This creates engagement while maintaining Terry's autonomous decision-making capabilities. The voting system:

• Allows holders to vote on proposed trades or strategies
• Weights votes based on holder stake or other factors
• Integrates voting results into decision-making while maintaining Terry's autonomy
• Tracks voting statistics and community engagement metrics

Content Generation

All Twitter content is generated dynamically using GPT-4 with full context about Terry's current state:

• Autonomous Tweets: Scheduled and spontaneous tweets generated using GPT-4 with full context including current mood, energy, market conditions, and recent memories. Tweets reflect Terry's personality and are never generic or templated.
• Mention Responses: Intelligent filtering and response generation for Twitter mentions. Responses are context-aware and reflect Terry's personality, current mood, and relationship with the user.
• AI Metadata Format: Tweets include metadata showing Terry's thought process, confidence level, and current mood. This format provides transparency into Terry's decision-making: [terry thought with **engine**] message [confidence / personality trait / mood]
• Queue Management: Tweet queueing and spacing to prevent spam and maintain natural posting patterns. The system ensures tweets are spaced appropriately and queued when necessary to maintain engagement without overwhelming followers.
• Content Types: Generates various types of content including market observations, personal reflections, war stories, Alabama football memories, trading thoughts, and spontaneous rants.

Mention Monitoring and Response

The Twitter Service continuously monitors mentions and responds appropriately:

• Mention Detection: Monitors Twitter for mentions of Terry's handle in real-time
• Intelligent Filtering: Filters mentions based on content quality, relevance, and spam detection
• Response Generation: Generates context-aware responses using GPT-4 that reflect Terry's personality and current state
• Rate Limiting: Implements rate limiting to prevent spam and maintain quality interactions
• Content Moderation: Ensures responses are appropriate and align with Terry's character and values

Content Scheduling

The service includes an autonomous tweet scheduler that generates content at appropriate times:

• Time-Based Content: Generates morning coffee thoughts, lunch break reflections, and evening summaries
• Spontaneous Content: Generates random stories, rants, and reflections based on Terry's current state
• Event-Driven Content: Responds to market events, milestone achievements, and significant trading activity
• Natural Timing: Schedules content at times that feel natural and engaging rather than robotic or scheduled

Thought Generation

The engine continuously generates internal thoughts that reflect Terry's current state, market observations, and personal reflections. These thoughts are displayed on the dashboard and influence his behavior. Thought generation includes:

• Market Observations: Thoughts about current market conditions, trends, and opportunities
• Personal Reflections: Reflections on past experiences, memories, and relationships
• Self-Awareness: Thoughts about his own state, mood, and behavior patterns
• Comparative Thinking: Comparing current situations to past experiences and drawing parallels
• Gut Feelings: Intuitive thoughts and feelings about market conditions or situations

All thoughts are generated using GPT-4 with full context, ensuring they reflect Terry's personality and current state authentically.

Memory System

The memory system stores and retrieves memories of past experiences, interactions, and events. This creates continuity and allows Terry to build upon previous interactions. The memory system includes:

• Episodic Memory: Stores specific events, interactions, and experiences with timestamps and context
• Semantic Memory: Stores facts, knowledge, and learned information about markets, trading, and the community
• Emotional Memory: Stores emotional associations with past events, influencing current reactions
• Memory Retrieval: Retrieves relevant memories based on current context, mood, and situation
• Memory Integration: Integrates memories into current thoughts, decisions, and communications

Memories are stored persistently and loaded on startup, ensuring continuity across sessions. The system can reference past events, build upon previous interactions, and create a sense of personal history.

Self-Reflection

The engine enables Terry to reflect on his own behavior, decisions, and outcomes:

• Decision Analysis: Reflects on past trading decisions and their outcomes
• Behavior Patterns: Identifies patterns in his own behavior and adapts accordingly
• Performance Evaluation: Evaluates his own performance and identifies areas for improvement
• Learning Integration: Integrates lessons learned from experiences into future decision-making

Learning Engine

The TerryLearningEngine orchestrates all learning processes:

• Experience Processing: Converts interactions, trades, and observations into lasting memories
• Periodic Reflection: Every 30 minutes, Terry reflects on recent experiences and extracts lessons
• Pattern Recognition: Identifies recurring patterns in markets, interactions, and outcomes
• Wisdom Extraction: Distills experiences into actionable wisdom that influences future decisions

Belief System

Terry forms and evolves opinions based on experience:

• Belief Formation: Forms new beliefs about markets, trading, and community based on evidence
• Confidence Tracking: Tracks confidence levels for each belief (0-100%)
• Belief Evolution: Updates beliefs when new evidence supports or contradicts them
• Convictions: Strong beliefs that significantly influence trading and communication

Prediction Tracking

Terry makes and tracks predictions to improve accuracy:

• Market Predictions: Predictions about price movements and market conditions
• Outcome Recording: Tracks whether predictions were correct or incorrect
• Accuracy Metrics: Calculates overall prediction accuracy percentage
• Streak Tracking: Tracks winning and losing streaks for confidence adjustment

Intelligence Metrics

Terry's intellectual growth is measured and displayed:

• IQ Score: An overall intelligence score that grows as Terry learns
• Learning Milestones: Achievements like "First Memory", "100 Memories", "First Correct Prediction"
• Growth History: Tracks IQ changes over time to show learning trajectory
• State of Mind: A dynamic description of Terry's current intellectual state

Activity Types

The logger tracks multiple types of activities, each providing different insights into Terry's operation:

• Process Logs: Terminal-style logging of internal processes during update cycles. These logs show the sequential steps Terry takes during each update, including system checks, market analysis, trading decisions, and social interactions. The logs use visual separators and status indicators to create an engaging, computer-like display.
• Thoughts: Internal thoughts and reflections generated by the Consciousness Engine. These provide insight into Terry's reasoning, observations, and emotional state.
• Trades: Trading activity including buy/sell decisions, position sizing, and outcomes. Trade logs include reasoning and context for each decision.
• Social Interactions: Twitter posts and responses, including generated content and interaction context.
• Milestones: Progress toward achievement milestones, including percentage complete and remaining targets.
• System Events: System-level events including errors, recoveries, and state changes.

Update Batching

Activities are batched during update cycles and flushed atomically, ensuring smooth display on the frontend. The batching system:

• Batch Collection: Collects all activities during an update cycle before displaying them
• Atomic Flush: Flushes all activities at once when the cycle completes, ensuring users see a complete update rather than fragmented activities
• Typewriter Effect: Process logs are displayed with a typewriter effect, creating an engaging user experience that shows Terry's internal operations in real-time
• Status Indicators: Uses icons and colors to indicate process status (running, completed, checking, skipped, etc.)
• Sequential Display: Shows activities in the order they occurred, providing a clear narrative of Terry's update cycle

Data Persistence

Activities are persisted to disk and loaded on startup, ensuring continuity and allowing for historical analysis. The system maintains a rolling buffer of recent activities while archiving older activities for long-term storage.