Plan Overview
This comprehensive Task-Agnostic Plan Framework provides a structured, reusable, and machine-readable blueprint for Meta-AI to tackle diverse high-level goals. It deconstructs complex objectives into abstract, actionable steps, orchestrates workflows, and defines clear, context-aware instructions for AI agents.
Modularity
Abstract steps can be reused across different tasks and domains
Scalability
Framework scales from simple to complex objectives
Machine-Readable
JSON Schema implementation enables programmatic execution
Workflow Structure
Define Goal
Clearly articulate and document the overarching objective or mission.
Deconstruct Objective
Break down the overarching objective into a sequence of logical, actionable sub-steps or phases.
Generalize Steps
Refine the deconstructed sub-steps to remove domain-specific details, jargon, and concrete examples.
Structure Workflow
Take the abstract steps and structure them into a coherent, logical workflow.
Design Plan Representation
Create a visual or structural representation of the plan.
Implement Plan Structure
Translate the designed representation into a functional, machine-readable format.
Refine Instructions and Structure
Review the implemented structure and refine the system prompts.
Final Abstraction Pass
Incorporate feedback to ensure steps are truly task-agnostic.
Finalize Documentation
Create the definitive version for deployment and use.
Define Goal
Clearly articulate and document the overarching objective or mission.
Deconstruct Objective
Break down the overarching objective into sub-steps.
Generalize Steps
Refine to remove domain-specific details.
Structure Workflow
Organize steps into a coherent workflow.
Design Plan Representation
Create visual/structural representation.
Implement Plan Structure
Translate design into machine-readable format.
Refine Instructions
Optimize prompts for clarity and context.
Final Abstraction Pass
Ensure steps are truly task-agnostic.
Finalize Documentation
Create definitive version for deployment.
JSON Schema Implementation
The machine-readable implementation of the Task-Agnostic Plan Framework as a JSON Schema:
Team Roles
TASe
Task-Agnostic Step extractor. Analyzes high-level objectives and breaks them down into fundamental, reusable steps.
Lyra
Prompt Engineer. Transforms abstract steps into structured workflows with clear system prompts and instructions.
Aurora
Designer. Creates intuitive visual/structural representations of the plan with focus on user experience.
Kodax
Engineer. Implements the plan structure in a functional, machine-readable format for programmatic execution.
