Ouroboros is an AI-driven framework designed to transform ambiguous human requirements into clear executable tasks. Central to its philosophy is the understanding that human input can often be irrational, incomplete, or contradictory. Through the application of Socratic Questioning and Ontological Analysis, Ouroboros iterates until a precise specification, termed a Seed, is achieved with minimal ambiguity. This process ensures that only valid inputs guide the execution of tasks, effectively replacing the pitfalls of Garbage In, Garbage Out (GIGO) with a robust methodology that fosters clarity and purpose.

Key Features

• Frugal by Default: Designs solutions with resource efficiency in mind, escalating only to more complex models when absolutely necessary.
• Rigorous in Verification: Ensures that critical evaluations undergo thorough scrutiny, avoiding rushed decisions that lack depth.
• Six Phases of Development: Incorporates a structured approach where tasks evolve through distinct phases, from Big Bang (initial conceptualization) to Secondary Loop (iterative refinement and enhancement).

The Six Phases

1. Big Bang: Establishes clarity from chaotic ideas through effective questioning.
2. PAL Router: Implements a Progressive Adaptive LLM model selection tailored to task complexity.
3. Double Diamond: Utilizes discover, define, design, and deliver principles to ensure thorough exploration and execution.
4. Resilience: Addresses stagnation by fostering lateral thinking and persona rotation for innovative solutions.
5. Evaluation: Employs a systematic approach through mechanical, semantic, and consensus evaluations.
6. Secondary Loop: Invokes a TODO registry for deferring non-essential tasks, focusing resources on primary objectives.

Economic Model

Ouroboros introduces a tiered cost structure that allows users to select the appropriate level of complexity for their tasks. This model optimizes costs significantly — approximately 85% reduction when compared to traditional methods that default to more sophisticated models without consideration for simplicity.

# Example of selecting an approach based on task complexity

def select_approach(task):
    if task.complexity < 0.4:
        return Tier.FRUGAL  # Start with a simple approach
    
    if task.reversible:
        return Tier.STANDARD  # Prioritize speed for reversible tasks
    
    if task.affects_direction:
        return Tier.FRONTIER  # Deep thinking for impactful decisions
    
    return Tier.STANDARD

Lateral Thinking Agents

To combat stagnation, Ouroboros employs diverse personas that switch perspectives when challenges arise, ensuring fresh approaches and renewed insights.

Ouroboros not only addresses the complexities of human input but also embodies its namesake — the eternal cycle of learning and growth. As the serpent that devours itself in cycles, the framework emphasizes the importance of iterative development and the wisdom derived from failure and success alike.

For detailed usage and implementation, please refer to the documentation and quick start guide included in the repository.